The Add date & time step adds a column with the current date and time, with customizable offset by days and timezones. The current date and time will be determined at the time the flow is run. This step can be useful for tasks like logging updated times.

Input/output

Connect any data into this step. It'll give you the output data of a new "Current DateTime" column appended to your dataset where the current date and time is noted (based on when the flow is run).

Default settings

By default, this step will have "Current DateTime" in the Column Name field, but you can customize this name to anything you prefer.

All date and time values created look like this format: 2019-09-18 12:33:09 (which is a format of YYYY-MM-DD hh:mm:ss). If you prefer a different format, connect a Format dates step right after this one to get the date values in your preferred way.

Custom settings

A date that is 0 days offset will be the current date and time. If you choose a positive value, it will be in the future, while a negative value will be in the past. If we add 1 to the Days Offset field, it shows us the date time stamp for this time tomorrow. If we add -1 to the Days Offset field, it'll show us the date time stamp for this time yesterday.

You can create multiple Add Date & Time Rules using a single step. Each rule will make a new column and the values of the column will be the same for every row. You might make multiple columns if you need the current date and time across different time zones or you want to have columns of varying offsets.

The Add growth rate step calculates the growth rate from the previous row within a column of your choosing.

Input/output

The data we'll connect into this step shows us how many items of a certain product we've sold, over the course of a few weeks.

After connecting input data to this step, it gives us the output of a newly added column called "Growth Rate" showing how much more or less of a product sold compared to the row (week) above it.

Custom settings

First, select the column that you'd like to calculate the growth rate of.

Then, under New Column Name field, type in the name you'd like this new column to be called. In the Decimal Places field, decide on how many decimal places you want included in the calculation and place this numeric value.

Finish by selecting the button Show Updated Results to save and display your new output data. Your table now has a column inserted right next to the one chosen for calculations. It will display the growth rate based off of the previous row. The results are percentages displayed in decimal format.

The Add if/else column step enables you to insert a conditional column filter.

Watch this Parabola University video to see the step at work.

A conditional statement is a core building block of any programming language. In Parabola, these are Add if/else column steps. The goal of each Add if/else column step is to create one column where the value of each cell in that column is dictated by the rules you create within the step.

Input/output

Our input data for this step is a table with four columns.  

Our output data, after using this Add if/else column step, has a new fourth column "Need to place an order?" that tells us whether to place an order and the values in "Type".

Custom settings

To enter an additional criteria, click into the + and select the column you'd like to run the criteria against. You can switch between adding condition filters by clicking on and and switching it to or.

To add an additional rule, click on the + Add a rule button.

If you'd like the step to reference a match between two columns, reference the column by placing them in curly brackets like so: {column name}.

If you'd instead like to search for specific values, you can enter the letter or number characters instead of the column name wrapped in curly braces.

Then, enter what you'd like the ew value to be in the set the new column value to section.

Finish by saving this step's settings by selecting the "Show Updated Results" button on the bottom of the lefthand-side toolbar.

Helpful tips

• The best use of this step includes (A) column name, (B) default value if no rules match, and (C) at least one rule.
• The "+ Add Rule" button adds another if statement. How it evaluates determines which value is set in the output. If it's true, it sets it to the new value. If it's false it moves to the next rule below it. Later rules cannot overwrite values set from earlier rules.
• Technically all you need to use this step and insert a conditional column matching rules you've created are: (A) a column name and (B) a default value if no rules match. With only a default value and no rules, the step will create a column with rows filled of your chosen value.

The Add math column step executes a math equation on each row of data. This step functions similarly to writing a formula in a new Excel column.

Check out this Parabola University video to see the Add math column step at work.

Input/output

Our input data has two rows of data: one for last week and one for this week, showing us how many Apples, Oranges, and Pears we sold.

After connecting the data into this step, it produces the output of a new column "Total Goods Sold" that shows calculations of the number of apples, oranges, and pears sold  last week and this week.

Custom settings

First, connect your flow's most recent step to this one and choose a New Column Name.

Then in the Calculation field, enter your formula. Be sure to wrap your column names in {curly braces}. As you write your formula, you'll notice that column names are selectable for auto-completion. This is handy for when you have large datasets with many columns.

If you see a the error below, your formula has an error. Check to make sure your headers are correctly entered, or that you're using the correct mathematical notation.

You can add as many calculations as you'd like by clicking Add Math Rules. For example, if you'd like to add a new "Yearly Forecast" column that predicts the Total Goods Sold in a year based on the weekly sold quantity, you can multiply your Total Goods Sold column by 52 weeks in a year to instantly generate those results.

Click Show Updated Results to save your selections and the step's settings are now complete! Your new columns will display your calculations on a row-by-row basis.

Helpful tips

• If you'd like to use a new column you created using the Add math column step in a previous formula, connect another Add math column step to this one.
• The Add math column step can handle positive and negative numbers with currency symbols, percentage symbols, decimals, and commas. There is no need to reformat these numbers in order to use them in this step. If your number is in the accounting format, you will need to use the Format numbers step to change the formatting before using it in this one.
• This step currently supports the Addition ( + ), Subtraction ( - ), Multiplication ( * ), Division ( / ), Modulo ( % ), and Power ( ^ ) operators
  

The Add row number step creates a new column adding sequential numbers to each row.

You can sort the numbers in ascending or descending order. You can add row numbers down the whole table, add numbers that repeat after a certain number of rows, repeat each number a set amount of times before advancing to the next number, and number based on another column. Numbering based on another column is a powerful way to add numbers to things like line items that share an order ID, or to rank your data.

Input/output

The input data we'll use with this step is a table of 15 rows with the 4 columns from a sales report.

The new output table has an additional column named "Row Number". The first cell in the column starts with 1 and increases in ascending order. The step will add a new column with sequential row numbers to any connected dataset.

Default settings

By default, the Column Name will be set to Row Number,  Start with will be set to 1, and the order will be ascending.

Custom settings

To sort the numbers in descending order, expand the "Numbering Settings" section and change the default setting from the top to  the bottom. The numbers will now be sorted from highest to lowest.

To see how many times a certain data type (like payment type) appears in your data set, you can change Number the entire table to Number based on a column such as "Name", "Product ID", and more.

There are also options to have this "Row number" column repeat the same integer a certain amount of times (Repeat numbers) or restart after a certain amount of rows (Restart after).

Helpful tips

• If two tables have the same number of rows but do not share a matching column, use this step and then the Combine tables step to create a new column that you can use to combine the tables.

The Add rows step adds any number of rows to a table. You can include default values for each column to make it easier to build a large table with duplicated values.

Input/output

This table has 8 rows with 5 columns.

To add more rows with default values, use the Add rows step. The output table has 5 additional rows. The "Time" column has a default value of 123, the "Apples" column has a default value of Parabola, the "Oranges" column has a default value of Labs, and the remaining columns are left blank

Custom settings

In the Number of Rows to Add field, enter the total number of new rows to add to a table.

In the "Column" Value (Optional) fields, set the default value to be added to each cell in a column. The value can be a string or a number.

The Add running total step calculates the running total sum of any column you choose.

Running total for the entire table

This setting creates a running total by summing the values in a specified column in every row. Each row will contain the sum of all of the previous values in the specified column plus the current row.

Running Total for each value in a column

Switch the first setting from “for the entire table” to “for each value in this column” to create individual running totals for each unique value in another reference column.

For example, if you needed to create the running total for the number of steps each person took in a day, you would sum the values in the “Steps” column and create a a running total for each value in the “Name” column.

The column being referenced for grouping does not need to be sorted, but the results will be easier to read if that column is sorted.

Helpful tips

• Any value that is not a number will be treated as a 0, including words and blanks.
• This step can sum: positive and negative numbers (-10), decimals (10.55), currencies ($50), percentages (10%), and scientific notation (10e2).
• Percentages will be summed as their numeric format. So 10% will be summed as 0.1

‍

The Add text column step adds a column filled with repeating row values. The rows can be filled with any integer/number, letter, special character, or value of your choosing. You can also reference other column values and combine them into this new column.

Check out this Parabola University video for a quick intro to the Add text column step.

(Please note: the Add text column step was previously called Insert text column and Text Merge.)

Input/output

Our input data example has two columns: "Webinar ID" and "Registrant name".

After using this step, our output data now has a third column called "Example Column" with the value "1" applied to its rows.

Custom settings

Enter the name of the new column you'd like to create in the New column name field. In the Column value field, fill in the repeating value you'd like for the rows. Click on Show Updated Results to save and display your changes.

Choose whether to turn on or off the checkbox Skip if column already exists to have this column added only if it doesn't already exist in your dataset. This is  useful when working with API steps where the column structure may change based on the available data. Similarly, if you have flows running on a schedule that pull in data at varying periods, this step can be used to create a static column of values so your flow doesn't stop running due to data temporarily not brought in.

Helpful tips

• If the column this step is adding already exists, then by default it won't be added to your dataset. This can be turned on and off. This is useful for preventing flow errors you set on a schedule, where the data pulled in varies at times.
• You can create multiple new columns with this step by going to the left-side toolbar and clicking add a column under the first outlined section.

• The Column value field supports Parabola's standard merge tags syntax, which is {curly braces} surrounding a column name. You can reference multiple columns using multiple cards (sections of newly-created columns) within a single step. The new column sections go from top-down, so a column created in the first card can be referenced in the second card.
• You can add bits of missing data into columns such as spaces, dashes, and other formatting characters.

The Average by group step calculates the average of all values in one or more columns. This step is similar to the AVERAGEIF function in Excel.

Input/output

Our input data has 30 rows of order information showing how many quantities of a particular food type was sold to a particular customer.

After connecting data to this step and setting it up, it gives us the output data of what the average quantity sold per order is, per unique food type.

Default settings

After connecting steps, this one will auto-select a quantity column (like "Quantity Sold") and a data type category to display quantity sold per row (like "Food Type").

Custom settings

If you'd like different columns to be chosen, you may choose the columns to include in your average calculation and choose which rows it is grouped by selecting them from the dropdown menus.

You can optionally apply this to only the selected columns or apply to all columns except. We'd recommend using the option that requires fewer column selections since this saves you time.

The Calculate score step scores each row based on specified criteria and puts the new score into a column.

Input/output

In this example, the data we'll input into this step has a list of multiple leads throughout a few companies.

We'll want to ensure we're contacting the best possible lead at each company, so we'll use this Calculate score step to score these leads in order to prioritize the best ones to contact.

This step's output is a new "Score" column of calculated scores per lead row.

Custom settings

After connecting our data into this step, in the Score Column Name field enter what you'd like the new column name to be.

To set up the step's rule(s), select the column that should have the rule applied to it, select a rule from the options provided, and type in a value this step should search for.

The rule options are: Blank, Not Blank, Equals, Not Equals, Contains, Not Contains, >, > or =, <, < or =.

Lastly, finish this rule setting by placing a number in the Increment Score field and selecting Show Updated Results to save it. In the above screenshot example, VP-titled people have a score of 3,  Director-titled people have a score of 2, and Manager-titled people have a score of 1.

Helpful tips:

• Add as many rules as you'd like by clicking on the Add Scoring Rule button.

The Categorize with AI transform step evaluates data sent to it and categorizes the rows based on categories that you predefine.

At launch, you can use Parabola AI steps at no extra charge to your team. After a beta period, we’ll move to a usage-based model for Parabola AI. (You will not be charged retroactively for usage during the beta period.)

Examples of categorizing data with AI

• Take a list of product names and categorize them by department: clothing, home goods, grocery, other
• Take a list of email addresses and sort them by type into these categories: work, personal, school, nonprofit, government, other
• Take a list of news headlines and categorize them by section: politics, business, world news, science, etc.

After running the step, it’s normal to modify your categories and re-run it as you see how the AI responds to your requests.

How to use this step

Selecting what to evaluate

You start by selecting which columns you want the AI to evaluate to produce a result.

• All columns: the AI looks at every data column to find and extract the item it’s looking for
• These columns: choose which column(s) the AI should try to extract data from
• All columns except: the AI looks at all columns except the ones you define

Note that even when the AI is looking at multiple (or all) columns, it’s still only evaluating and generating a result per row.

Setting the categories

This step is designed to assign categories to rows, so it needs to know what the desired categories are. The step provides spaces to write in as many categories as you need (it starts with three blank ones only as an example). Add one category per space.

Fine tuning

Open the "Fine tuning" drawer to see extra configuration options. Using this field, you can provide additional context or explanation to help the AI deliver the result you want.

For example, if you gave this step a list of animals and asked it to categorize them as “Animal I like” vs. “Animal I don’t like,” it might not give you an accurate result! But you could then use this field to say:

“I tend to like furry animals that are friendly to humans, like dogs and horses and dolphins, and not others”

This step would then better understand the categorization you’re looking for.

Helpful tips

• Currently, the AI can only run a few thousand rows at once. Choose and trim your data accordingly
• Sometimes you’ll see a response or error back instead of a result. Those responses are often generated by the AI, and can help you modify the prompt to get what you need.
• Still having trouble getting the response you expect? Often, adding more context in the "Fine tuning" section solves the problem.

‍

With our Artificial Intelligence (AI) steps, Parabola lets you process your data with OpenAI’s GPT AI in specific, useful, and reliable ways. But working with AI comes with important considerations.

AI has natural limitations

AI is a new field in technology, and while the results are sometimes exciting, they’re often less dependable than traditional human-built data processes. Consider reading OpenAI’s breakdown of their AI’s limitations, and keep the following in mind when using an AI to process data:

1. Model limitations: Understand GPT's knowledge cutoff. GPT can “hallucinate” or be confidently incorrect, so do not expect results to be perfectly accurate all of the time.
2. Data sharing: When data is processed using Parabola’s AI steps, that data is sent to OpenAI, a 3rd party. Review their policies and practices to understand how they handle your data.
3. Monitoring: Continuously assess GPT performance; take corrective actions as needed.
4. Responsible use: Adhere to regulations; inform stakeholders of limitations and risks.

AI processing is … different

We’ve made data processing with an AI easier than ever before! But when you use AI as part of a Parabola Flow, those steps can be less transparent and reliable than the rules-based transform steps that your Parabola Flows normally use.

Keep this in mind especially when automating processes where exact precision is critical, like financial data. Consider using AI for steps that require “interpretation” — which AI can be quite good at! — rather than precise calculation.

Feedback feeds us!

If you have feedback about the usefulness of these steps, or the AI-generated responses you’re getting from them, please tell us!

‍

The Change text casing step converts the text in any column to a selected case type.

Input/output

Our input data is customer information that displays first and last names in all uppercase. We can use the Change text casing step to turn our first and last names into title case.

As seen in our output data below, this step made our "first_name" and "last_name" columns change into title case.

Custom settings

The first thing to do when you connect data to the Change text casing step is to select the column(s) you'd like to Recase. You can select as many columns as you need.

Then, you will select the desired Case Type. The available case types are:

• Upper Case (THIS IS UPPER CASE)
• Lower Case (this is lower case)
• Title Case (This is Title Case)
• Sentence Case (This is sentence case)
• Proper Case (This Is Proper Case)
• Snake Case (this_is_snake_case)
• Camel Case (thisIsCamelCase)
• Train Case (This-Is-Train-Case)
• Kebab Case (this-is-kebab-case)

Helpful tips

• This step can only change one case type at a time. If you'd like to change the casing of your columns to different case types, you can do this by connecting a chain of multiple Change text casing steps one after the other.

The Clean data step removes leading or trailing spaces and other unwanted characters (letters, numbers, or punctuation) from any rows of data you'd like.

Input/output

Our input data has a column of messages with various leading and trailing spaces.

By using the Clean data step, we can easily remove spaces to clean up the "Message" column.

Default settings

When you first connect data into this step, by default your table's first column will be auto-selected to clean and the option to Remove all spaces will be applied.

Custom settings

To customize these settings, first choose the column(s) from the dropdown that the cleaning rules should apply to. You can either select to Clean the data in these columns or Clean the data in all columns except.

Next, choose the cleaning options that should be applied to the specified column(s). The available options are:

• Remove all spaces
• Remove leading spaces
• Remove trailing spaces
• Remove all punctuations
• Remove all characters
• Remove all numbers
• Remove all non-numbers
• Clean to use as JSON

You can combine the cleaning options if needed.

You can also add multiple clean data rules to this step by clicking on the Add Rule button.

In our above screenshots, we selected the "Message" column to apply our cleaning rules to and selected two cleaning options: Remove leading spaces and Remove trailing spaces. As seen in the screenshot below, this cleaned up varying leading and trailing spaces that were in the "Message" column's rows.

Helpful tips

• This step is useful to use when you're preparing data to send to an API using the Send to API step. If you encounter a JSON error when sending data to an API, try using the Clean data step first to see if it can quickly remove the special character(s) causing the issue.

The Combine Tables step joins multiple tables into one by matching rows between those tables. It is similar to a VLOOKUP in Excel or Google Sheets. In SQL, it's like a join clause.

The principle is simple: if we have two tables of data that are related to each other, we can use the Combine Tables step to join them into one table by matching rows between those tables.

This step can handle combining two tables at a time. Once we set it up, we can use it repeatedly. Even if the row amount changes, our step will continue working.

Check out the Parabola University video below to learn more about the Combine Tables step.

Input/output

The Combine Tables step requires two data inputs to combine. In our example below, we have two tables feeding into it. The first one is a table of Variant IDs, Product Name, Variant Name, and Sale Price. The second one is a table of Variant IDs and Stock Quantity.

After using the Combine Tables step, our output data (shown below) has combined the "Stock Quantity" column from the second table to the first table using "Variant ID" as the matching identifier.

Custom settings

After connecting two datasets into this step, in the left-side toolbar choose whether you will Keep all rows or Keep only matching rows for your data sources using the drop down menus.

Then in the section above the button + add rule, click select to access the dropdown menu of columns and select which ones to join the tables by. The column you choose should be present in both datasets, have identical values, and be used as the unique key identifier between them.

Once you're done, select the button Show Updated Results to save and display the newly-combined tables.

If the datasets you want to combine don't have a shared column of identical values, then you can use the step Insert row numbers to make one in each table. You'd do this by connecting your import datasets to an Insert row numbers step for generating a column with incrementally-increasing numbers.

‍

Other ways to combine tables:

Now we've learned one example using our default settings. Let's explore the other ways that we ca combine tables in Parabola. We can switch our Combine tables step to keep all rows that have matches in all tables. The resulting table will only contain rows that exist in every table we are combining.

The last option is to keep all rows in all tables. This will create a resulting table that has every row from every table present. If no match was found for a specific row, it will still exist. This option has a tendency to create a lot of blank cells, and can be tricky to use properly.

‍

Helpful tips

• The default setting for the Combine Tables step is the most common way to combine tables. It keeps the entire primary table and finds matches in the other tables, fitting them in to their matched rows as we go. If a row doesn't have a match in the primary table, it won't show up in the results.
• To combine three or more tables together, chain together multiple Combine Tables steps to merge your data.

• If we need multiple rules to determine a match, by default we'll determine row matches if any rules apply. We can change this so that rows match only when all of the rules apply. Notice that the word "or" changes to "and" when we update this setting.

Troubleshooting tips

Here are a few common questions that come up around the Combine tables steps and ways to troubleshoot them:

Issue: I’m seeing my rows multiplied in the results section of the Combine tables step.

Solution: It’s possible that you are matching on data that is duplicated in one or both of your sources. Check your inputs to make sure the selected matching column have unique data in both inputs. If you don’t have a column with unique data, you may need to choose multiple columns to match on to make sure you’re not duplicating your rows.

Here is a quick video walking through this process:

Issue: I’m seeing a Missing columns error in the Combine tables step.

Solution: This can happen when you make changes to one of your inputs after you’ve already selected the setting in your Combine tables step. You’ll need to either reselect your matching columns in the step or pull in a new Combine tables step.

The Compare dates step compares dates in a column to another column of dates, or compares dates in a column against the current time when your Parabola flow runs. This step is similar to the DATEDIF function in Excel by creating a new column showing the measurement of time between two dates.

Input/output

The data you connect to this step needs at least one column with date information. In this example, we pass a column called "Sales Date" and compare that date against Current time. The step provides an output of a column auto-named "Difference" showing us the number of days between "Sales Date" and Current time.

Default settings

By default, this step will detect a column with the word "date" in its header, and find the number of Hours between values in this column and Current time, expressing the results in a new column named Difference. In this example, the step auto-selected the column "Sales Date".

Custom settings

Select which time unit measurement you'd like to perform calculations with.

The weekdays option will calculate the number of days, excluding weekends, between the two dates, akin to the NETWORKDAYS function in Excel.

If your data has a time zone different from your current location's, you can indicate that in the step. You can find the time zone selection by clicking the three-dot button under Date 1 or Date 2

If you would like to compare multiple columns of dates data, you may select another column in the table instead of the default Current time option.

Helpful tips

• Date formatting: we suggest formatting your dates either as YYYY-MM-DD HH:mm:ss or X (Unix) for this step. You can use the Format dates step to get your date values in one of these formats.
• Working with two-digit years (YY): by default, we determine the century by using a cutoff year of 2029. That means that two-digit years between “00-29” will be interpreted as the years between 2000-2029, and two-digit years between “30-99" will be interpreted as the years between 1930-1999.
• If any of the values being compared can't be recognized as a date or date-time, then the result will be blank for that row. Otherwise the output is always a number.

The Count by group step counts the number of rows that exist for each unique value in one or more columns. This step is similar to the COUNTIF function in Excel.

Watch this Parabola University video for a quick walkthrough of the Count by group step.

Input/output

Our input data is a list of product variants.

We want to know how many product variants we have available at different price points. After connecting the data to this step and setting it up, it gives us the output data of the unique variants amount per unique sale price.

Default settings

By default, this step will count every row as unique and display a "Count" column with the same value as the number of rows in your input table.

Custom settings

To customize these default settings, you'll first select a column or multiple columns you'd like to count unique values for. Then, you'll provide a name for that new column that will display the count of unique values.

In the example above, we wanted tocount unique variants per unique sale price, so we selected Sale Price from the dropdown and named the new column "Variant Count".

We can add multiple columns if we want to count unique values for the combinations of those column values.

To build on the example abve, if we want to count unique variants per quantity sold, we can select both of those columns from the dropdown to see these updated results. The screenshot below shows us that I have 4 variants of "Tees" priced at $25 available in our store.

‍

The Edit columns step allows you to keep or remove columns, rename them, and then reorder those columns – all from one step! This step will replace the following steps: Select columns, Rename columns, and Reorder columns.

Check out this Parabola University video to see the Edit columns step in action.

Keep or Remove columns

When you first connect your data into the Edit columns step, the Keep/Remove section will auto-select to Keep all columns. From the dropdown you will have 3 options:

• Keep all columns
• Keep these columns
• Remove these columns

If you choose to Keep all columns, you will see all of the columns in your data present.

Choosing to Keep these columns, will allow you to select the columns that you would like to keep. You can use the Add all or Clear all button if you need to bulk update or clear your selection:

The Remove these columns option will remove the columns that you are selecting from your data. There is an Add all and Clear all option that you can use to add all of the columns or clear your selection:

There is an option to Guarantee your kept columns. Guaranteed columns are kept even if they are later missing from the input data (e.g., an API that doesn’t include a column of data if it’s blank). Check the box to keep your columns even if they are later missing from the data you are pulling in:

Rename and Reorder columns

From the Rename/Reorder section of the step, you can update the names of your columns. To do this, click the name of the column in that section. You should then be able to type in or update the column’s name. If you need to reset the name of the columns, you can do so individually by clicking the reset icon next to the column or reset all the names by clicking the Reset names button.

To update the order of your columns, click and drag the handles to the left of the column positions and drag into place. Alternatively, you can click the position number and type in a new number. You can reset the column order by clicking “Reset order.”

By default, any new columns will be placed at the end of your table of data. If you disable this checkbox, only columns that you have explicitly moved to new positions will be reordered, and all other columns will attempt to position themselves as close to their input position as possible.

When this checkbox is disabled, any column that has been moved into a specific position will show a bolder drag handle icon and position number.

Disabling this setting is best used when you know you need specific columns in specific numerical positions, and do not want to reorder the other columns.

If you have an Edit Columns step that existed prior to this change, disabling the checkbox will show all columns with bold positions, indicating that they are all in set locations. To fully take advantage of this feature with an existing Edit Columns step, disable the checkbox and click “reset order”. Resetting this step will allow you to place columns in exact positions.

Use the Enrich with API step to make API requests using a list of data, enriching each row with data from an external API endpoint.

Input/output

Our input data has two columns: "data.id" and "data.employee_name".

Our output data, after using this step, has three new columns appended to it: "api.status", "api.data.id", and "api.data.employee_name". This data was appended to each row that made the call to the API.

Custom settings

First, decide if your data needs a GET or POST operation, or the less common PUT or PATCH, and select it in the Type dropdown. A GET operation is the most common way to request data from an API. A POST is another way to request data, though it is more commonly used to make changes, like adding a new user to a table. PUT and PATCH make updates to data, and sometimes return a new value that can be useful.

Insert your API Endpoint URL in the text field.

Sending a body in your API request

• A GET cannot send a Body in its request. A POST can send a Body in its request. In Parabola, the Body of the request will always be sent in JSON.
• Simple JSON looks like this:

{ "key1":"value1", "key2":"value2", "key3":"value3" }

Using merge tags

• Merge tags can be added to the API Endpoint URL or the Body of a request. For example, if you have a column named "data.id", you could use it in the API Endpoint URL by including {data.id} in it. Your URL would look like this:

http://third-party-api-goes-here.com/users/{data.id}

• Similarly, you can add merge tags to the body.

{
"key1": "{data.id}",
"key2": "{data.employee_name}",
"key3": "{Type}"
}

• For this GET example, your API Endpoint URL will require an ID or some sort of unique identifier required by the API to match your data request with the data available. Append that ID column to your API Endpoint URL. In this case, we use {data.id}.
• Important Note: If the column referenced in the API Endpoint URL is named "api", the enrichment step will remove the column after the calculation. Use the Rename Columns step to change the column name to anything besides "api", such as "api.id".

‍

Authentication

Most APIs require authentication to access their data. This is likely the first part of their documentation. Try searching for the word Authentication in their documentation.

Here are the Authentication types available in Parabola:

The most common types of authentication are Bearer Tokens, Username/Password (also referred to as Basic), and OAuth2.0. Parabola has integrated these authentication types directly into this step.

Bearer Token

This method requires you to send your API Key or API Token as a Bearer Token. Take a look at this example below:

The part that indicates it is a bearer token is this:

-H "Authorization: Bearer sk_test_WiyegCaE6iGr8eSucOHitqFF"

To add this specific token in Parabola, select Bearer Token from the Authorization menu and add "sk_test_WiyegCaE6iGr8eSucOHitqFF" as the value.

Username/Password (Basic)

This method is also referred to as Basic Authorization or simply Basic. Most often, the username and password used to sign into the service can be entered here.

However, some APIs require an API key to be used as a username, password, or both. If that's the case, Insert the API key into the respective field noted in the documentation.

The example below demonstrates how to connect to Stripe's API using the Basic Authorization method.

The Endpoint URL shows a request being made to a resource called customers.  The authorization type can be identified as Basic for two reasons:

1. The -u indicates Basic Authorizationusername.
2. Most APIs reference the username and password formatted as username:password. There is a colon : . This indicates that only a username is required for authentication.

To authorize this API in Parabola, fill in the fields below:

OAuth2.0

This method is an authorization protocol that allows users to sign into a platform using a third-party account. OAuth2.0 allows a user to selectively grant access for various applications they may want to use.

Authenticating via OAuth2.0 does require more time to configure. For more details on how to authorize using this method, read our guide Using OAuth2.0 in Parabola.

Expiring Access Token

Some APIs will require users to generate access tokens that have short expirations. Generally, any token that expires in less than 1 day is considered to be "short-lived" and may be using this type of authentication. This type of authentication in Parabola serves a grouping of related authentication styles that generally follow the same pattern.

One very specific type of authentication that is served by this option in Parabola is called OAuth2.0 Client Credentials. This differs from our standard OAuth2.0 support, which is built specifically for OAuth2.0 Authorization Code. Both Client Credentials and Authorization Code are part of the OAuth2.0 spec, but represent different Grant Types.

Authenticating with the Expiring Access Token option is more complex than options like Bearer Token, but less complex than OAuth2.0. For more details on how to use this option, read our guide Using Expiring Access Tokens in Parabola.

‍

The Expand JSON step converts JSON into a spreadsheet table format.

JSON stands for "JavaScript Object Notation" and is a widely-used data format. If you're using APIs, you've likely come across JSON and may want to expand it to access the data nested inside it.

If you're trying to troubleshoot a JSON expansion issue, reference this community post.

Input/output

In the input data below, it shows what came from a GET call made to a free API: https://api.punkapi.com/v2/beers. This data has multiple JSON columns with a table of 25 rows and 21 columns.

Below are one of the "method" column's row values formatted for easier reading. This is in the row where the column "name" is equal to "Berliner Weisse With Yuzu - B-Sides".

Once we use the Expand JSON step, our output data becomes 28 rows and 30 columns. We've gained 3 rows and 9 columns after expanding the "method mash_temp" dataset with all non-datasets also expanded.

If we filter for beers with the name "Berliner Weisse With Yuzu - B-Sides" and focus on the newly-expanded columns, we see that there's one row per element in the array "method mash_temp" for a total of 4 rows with "Berliner Weisse With Yuzu - B-Sides". If we look closely at the 6 additional columns that contain the newly-expanded data, we'll see the data nested within the JSON object above neatly arranged in a table format.

Default settings

By default, this step will try and convert as much JSON as it can, and in most cases we won't need to change settings. This step defaults to expanding the first JSON dataset it comes across.

By default, this step also sets to Expand all non-datasets. That setting is nested under the "Non-dataset Options" options. This means that all the object keys will be expanded out to as many columns as it requires.

If you're familiar with JSON, you can think of this step's functionality like the following: datasets are JSON arrays that are expanded into new rows. Non-datasets are JSON objects that are expanded into new columns.

Custom settings

To customize the default settings, we'll first want to make sure the right dataset is selected to be expanded. We can only choose one dataset to expand or can select to expand no datasets. Datasets are expanded so that each entry is put into a new row. If our data has no dataset, then we'll see N/A selected in the first option for Expand the N/A dataset.

Next, if we don't need all non-datasets expanded into columns, we can open the "Non-dataset Options" and switch it from Expand all non-datasets to Expand these non-datasets.

If we select Expand these non-datasets, we'll see an option to specify exactly which non-datasets we should expand into new columns. We can select as many as we require. Whichever non-datasets we choose will be expanded into new columns, and if there are more non-datasets contained within, those will be expanded as well.

‍

Helpful Tips

• If you have multiple rows of JSON and the data is not fully expanding as expecting, review the JSON body in the top row of your data. Parabola will look at the JSON in the first row and use that as a ‘template’ for expanding JSON in subsequent rows. If there is an invalid or missing key in the top row, our expand step JSON step will omit that field in subsequent expanded rows. For additional information on troubleshooting this issue, please reference this community post.

This step is simply a text box that lets you ask the AI to revise, remove, or add to the input data however you like. It’s the most flexible of the AI steps … but that means it has the most variable results, too!

At launch, you can use Parabola AI steps at no extra charge to your team. After a beta period, we’ll move to a usage-based model for Parabola AI. (You will not be charged retroactively for usage during the beta period.)

Examples of experimenting with AI using this step

• Take a list of product categories, and ask the step to assign emojis to each one
• Change the values in a column to Title Case … but only if the word looks like a name or a proper noun
• Analyze the sentiment — positive, negative, neutral — of a column with text
• Remove values from a column that meet a certain condition

How to use this step

This step is simply a text box, where you can make a request of the AI (also known as a “prompt”). Your results will depend in part on the specificity of your prompt. Here are some tips for making the most of this step.

• You can refer to existing columns by name. If it helps, you can even try putting them into “quotes” to isolate them.
• Prompts can be several sentences long. If you start out simply and don’t get the result you want, try explaining your ask in more detail!
• Sometimes using an example helps. E.g., if you’re trying to get the AI to rate the cuteness of animals, you can say “the cutest animals are fuzzy, like squirrels, rabbits, and pandas. young animals are also cute, like puppies, kittens, and bunnies”
• Check out OpenAI’s tips for writing a good prompt (aka “Prompt Engineering”)

Helpful tips

• Currently, the AI can only run a few thousand rows at once. Choose and trim your data accordingly
• Sometimes you’ll see a response or error back instead of a result. Those responses are often generated by the AI, and can help you modify the prompt to get what you need.

‍

With our Artificial Intelligence (AI) steps, Parabola lets you process your data with OpenAI’s GPT AI in specific, useful, and reliable ways. But working with AI comes with important considerations.

AI has natural limitations

AI is a new field in technology, and while the results are sometimes exciting, they’re often less dependable than traditional human-built data processes. Consider reading OpenAI’s breakdown of their AI’s limitations, and keep the following in mind when using an AI to process data:

1. Model limitations: Understand GPT's knowledge cutoff. GPT can “hallucinate” or be confidently incorrect, so do not expect results to be perfectly accurate all of the time.
2. Data sharing: When data is processed using Parabola’s AI steps, that data is sent to OpenAI, a 3rd party. Review their policies and practices to understand how they handle your data.
3. Monitoring: Continuously assess GPT performance; take corrective actions as needed.
4. Responsible use: Adhere to regulations; inform stakeholders of limitations and risks.

AI processing is … different

We’ve made data processing with an AI easier than ever before! But when you use AI as part of a Parabola Flow, those steps can be less transparent and reliable than the rules-based transform steps that your Parabola Flows normally use.

Keep this in mind especially when automating processes where exact precision is critical, like financial data. Consider using AI for steps that require “interpretation” — which AI can be quite good at! — rather than precise calculation.

Feedback feeds us!

If you have feedback about the usefulness of these steps, or the AI-generated responses you’re getting from them, please tell us!

‍

The Extract text step extracts a portion of text based on a matching character or offset. You may use this to pull out company names from emails, remove part of an ID, or extract the timezone from any date/time stamp.

Input/output

In this case, we're looking to extract email domains from our customer email addresses. Our input data has three columns: "first_name", "last_name", and "email.

After using the Extract text step, it gives us a new column named "email domain" where extracted email domains are listed (taken from the "email" column's values). This new column is filled with company names we may want to prioritize.

Custom settings

First, select the column that you'd like to extract text from.

Then, give your new column a name. This step will always create a new column with your extracted data.

Next, select an Operation. The options are:

• Find all text after
• Find all text after the chosen matching text or offset
• Find all text before
• Find all text before the chosen matching text or offset
• Find some text after
• Find a set length of text after the chosen matching text or offset
• Find some text before
• Find a set length of text before the chosen matching text or offset

Finally, you'll select the Matching Text or Offset. The options are:

• First instance of matching text: You'll set the Matching Text to look for and we'll delimit on the first instance.

• Last instance of matching text:You'll set the Matching Text to look for and we'll delimit on the last instance.

• Offset from beginning of text: You'll set the Offset Length and we'll count out that number of characters from the beginning of your text to determine the delimiter.

• Offset from end of text: You'll set the Offset Length and we'll count out that number of characters from the end of your text to determine the delimiter.

With any chosen matching text or offset option, you'll also be able to set a Max Length of Text to Keep.

The Extract with AI transform step evaluates data sent to it, then uses the GPT AI to extract whatever pieces of information you need. By naming the new columns that you’d like to populate, you can tell this step which pieces of information you’d like to extract.

At launch, you can use Parabola AI steps at no extra charge to your team. After a beta period, we’ll move to a usage-based model for Parabola AI. (You will not be charged retroactively for usage during the beta period.)

‍

Examples of extracting data with AI

• Processing a list of invoices and extracting the invoice amount, due date, sender, and more
• Taking a list of email addresses and extracting the domain (e.g., gmail.com)
• Taking inbound emails and extracting the sender, company, and request type

As you can tell from some of these examples, the AI can do some lightweight interpretation (e.g., naming a company from an email domain URL) as well as simple data extraction.

How to use this step

Selecting what to evaluate

You start by selecting which columns you want the AI to evaluate to produce a result.

• All columns: the AI looks at every data column to find and extract the item it’s looking for
• These columns: choose which column(s) the AI should try to extract data from
• All columns except: the AI looks at all columns except the ones you define

Note that even when the AI is looking at multiple (or all) columns, it’s still only evaluating and generating a result per row.

Identifying what to extract

The next part of this step serves two purposes simultaneously:

1. Telling the AI what items you’d like to extract from the input data (e.g., “full name”)
2. Naming the new column(s) that the extracted data will go into (e.g., a column named “full name”)

The step starts out with three blank spaces (as an example); you can fill those and even add additional columns to extract data to. Don’t need three? The step will automatically remove any blank ones, or you can remove them yourself.

(You can always rename or trim these columns later using other Parabola steps.)

Fine tuning

Open the "Fine tuning" drawer to see extra configuration options. Using this field, you can provide additional context or explanation to help the AI deliver the result you want.

For example, if the AI was having trouble pulling “Invoice number” from imported invoice data, you might explain to it:

“our invoice numbers tend to begin in 96 and are 12-15 digits long”

The AI would then better understand what you want to extract.

Helpful tips

• Currently, the AI can only run a few thousand rows at once. Choose and trim your data accordingly
• Sometimes you’ll see a response or error back instead of a result. Those responses are often generated by the AI, and can help you modify the prompt to get what you need.
• Still having trouble getting the response you expect? Often, adding more context in the "Fine tuning" section fixes the problem.

With our Artificial Intelligence (AI) steps, Parabola lets you process your data with OpenAI’s GPT AI in specific, useful, and reliable ways. But working with AI comes with important considerations.

AI has natural limitations

AI is a new field in technology, and while the results are sometimes exciting, they’re often less dependable than traditional human-built data processes. Consider reading OpenAI’s breakdown of their AI’s limitations, and keep the following in mind when using an AI to process data:

1. Model limitations: Understand GPT's knowledge cutoff. GPT can “hallucinate” or be confidently incorrect, so do not expect results to be perfectly accurate all of the time.
2. Data sharing: When data is processed using Parabola’s AI steps, that data is sent to OpenAI, a 3rd party. Review their policies and practices to understand how they handle your data.
3. Monitoring: Continuously assess GPT performance; take corrective actions as needed.
4. Responsible use: Adhere to regulations; inform stakeholders of limitations and risks.

AI processing is … different

We’ve made data processing with an AI easier than ever before! But when you use AI as part of a Parabola Flow, those steps can be less transparent and reliable than the rules-based transform steps that your Parabola Flows normally use.

Keep this in mind especially when automating processes where exact precision is critical, like financial data. Consider using AI for steps that require “interpretation” — which AI can be quite good at! — rather than precise calculation.

Feedback feeds us!

If you have feedback about the usefulness of these steps, or the AI-generated responses you’re getting from them, please tell us!

‍

The Fill addresses step creates a column with Google Maps address information and standardizes the format of your addresses, completes partial addresses, finds zip codes, or finds addresses for landmarks to businesses.

Input/output

Data connected to this step works best when it is in the shape of a column filled with place names. Think of your data as Google Maps search strings. In our example below, we have a table with a single column called "Landmarks," containing names of a few global destinations.

The resulting output, after using this Fill addresses step, is a new column autofilled with location information from Google Maps.

Custom settings

In the Address Column dropdown, select the column that has the location data you'd like Google Maps to search.

Next, in the Convert to dropdown, select how you'd like Google Maps to return the data. You can choose between the Full Address (if applicable) or simply the Zip Code.

If you check the Default to Original box and there are no address details found in Google Maps, your new column will return the data from your Address Column. If you'd rather leave the result blank where no address is found, then you may keep the box unchecked.

Lastly, give your new column a name in the New Column Name field.

Helpful tips

• This step is only available on our Advanced plan.
• This step will only run a maximum of 1,000 rows per step.

The Fill in blanks step uses the nearest value above, below, to the left, or to the right of a cell in a column. You can also use a custom value or reference another column's value to fill in the blank cell.

Input/output

We'll use the below input table for this step to process. It contains four columns including one with some blank rows called "Subsidiary".

After using the Fill in blanks step, our output table now has values in the previously-empty "Subsidiary" column. The Fill in Blanks step has populated the blank rows with the values to the left of that row cell (i.e. values in the left-side "Organization" column).

Custom settings

Once you connect data to the Fill in blanks step, you will first need to select which column we should fill in blanks for.

By default, this step will fill in blanks with values from above. To change this, click on values from above and select another option from the available dropdown:

• values from above
• values from below
• values from left
• values from right
• custom values

You can create multiple fill in blanks rules that will be applied from the top down. To do so, simply click on the blue link to Add Rule.

The Filter rows step keeps rows that satisfy a set of rules. You can use this step to create simple filters with one rule, or complex filters with many rules.

Check out this Parabola University video to see the Filter rows step at work.

Input/output

The input data we'll connect to this step has 15 rows of webinar registrants.

Let's say we want to filter this for people who registered after February 1, 2022. By setting a rule to include rows where the value in "Registration Time" column is greater than or equal to "2022-02-01 00:00:00", we immediately get nine rows of data showing who registered after February 1, 2022.

Custom settings

By default, this step will keep rows only if they follow any of the rules. This means that if a row satisfies one of the rules, that row will be kept. You can change this to keep rows only if they match all of the rules by selecting +add rule and changing OR to AND between them.

Now it's time to set up filters for rows. For each rule, you'll select a column, a filtering condition, and a matching value.

Available filtering conditions:

• is blank
• is not blank
• is equal to
• is not equal to
• contains
• does not contain
• is greater than
• is greater than or equal to
• is less than
• is less than or equal to

To add another rule, click on the blue + add rule button to configure another rule.

The contains, does not contain, is equal to, and is not equal to filter operations allows multiple criteria fields.

You can compare columns inside of your Filter rows steps by entering the column name wrapped in {curly braces}.

Helpful tips

Keeping rows between two values: If you need to only keep rows in a column whose values are between two values, then you'd use two separate rules. Let's say I want to keep values in a column that are between 5 and 10, including 5 and 10. I would set one rule to keep all rows that are greater than or equal to 5. I would set a second rule, select the AND option, and set to are less than or equal to 10.

Filtering dates: The Filter rows step can generally handle filtering for dates that match a certain criteria without any additional date formatting required. However, if you're experiencing any issues with filtering dates, we recommend trying to convert your dates into a Unix format or Lexicographical format. To update your date formats, use a Format dates step before sending your data to the Filter rows step. For Unix format, use capital X for precision to the second and lowercase x for precision to the millisecond. For Lexicographical format, use YYYY-MM-DD HH:mm:ss.

Accounting for text casing: Please note that rules are not case sensitive.

Reversing the filtering logic: If you set up a filter with multiple rules and wish to instead see the rows that are being removed by the filter, you can reverse your filtering by changing Keep rows to Remove rows.

Filtering symmetrically: If you need to split your table to create separate branches that will eventually be merged back together to recreate the complete data set, you will use a few Filter rows step while making sure you're not duplicating or losing any rows in that process. For example, let's say you have data that spans three months (October, November, and December) and you need to process each month differently. You want one Filter rows step per month. You'll need to use three separate Filter rows step with the following rules:

1. Date is less than 2019-11-01.
2. Date is greater than or equal to 2019-11-01 AND less than 2019-12-01.
3. Date is greater than or equal to 2019-12-01.

A good way to make sure you successfully filtered symmetrically is to use a Stack tables step at the end of your separate branches to merge the data back together. The resulting row count should match the row count of your starting data before the Filter rows steps.

The Find and replace step works like you might imagine the name means: You can type in a value, specify the columns to look for it in, and type in a new value to replace it with if it is found.

Input/output

Our input data below has orders information telling us how many units were sold today per product variant type.

Let's say we need to replace the blank values in the "Units sold" column with a "0". As seen below, once we connect this input data into the step, it produces that output where any blank cells in the column "Units sold" now has a "0".

Custom settings

To set up your Find and Replace rule, you'll need to:

1. Type in a value to look for. This input can be left blank if you'd like this step to look for blank or empty row cells.
2. Select the column(s) where we should look. By default, this step will search for the value in "all columns".
3. Type in a value to replace the value with if it's located. This input can be left blank if you'd like this step to replace the located values with blanks (i.e. replacing it with empty row cells).

You can create as many rules as you need by clicking on the + add rule button on the left-side step settings bar. The rules are applied in the order they appear from top to bottom. The first rule will be applied to every row and the necessary replacements will be made. Once that's completed, the next rule will be applied.

Helpful tips

• The Find and replace step ignores text casing. It will find matching values even if the text casing, such as capitalized letters, aren't an identical match.

The Find max by group step calculates the maximum value in a column or a group of columns. This step is similar to the MAX function in Excel.

Input/Output

Our input data has 11 rows of order information showing us how many items were sold, the food type, the customer name, and their payment type.

If we instruct the step to find the maximum value in the Quantity Sold column, and create one new row per value in the Food Type column, we can easily find the largest sale of each item.

Custom settings

This step has no default behavior, and must be customized to return a useful output.

First, decide whether you would like to include or exclude given columns by selecting the following columns or all columns except for.

Next, click the plus sign to choose the columns you wish to include or exclude.

If the Create a new row... option is left blank, the step will return one value per column, according to your specifications in Find the maximum...

To find maximum values for multiple attributes, you can Create a new row for each unique value in your desired columns. As before, you can include or exclude given columns, and you can click the plus sign to specify.

Unique value columns will add left to right, and sort A-Z, and your desired maximum value will display as a new column with "(max)" appended to its header.

The Find min/max step calculates the minimum or maximum value across a range of columns.

Input/output

Our input data shows how many items we sold on a given day.

We can use this step to produce an output with a new column at the right side, called "Most Sold in a Day," which contains the maximum value in each row.

Default settings

This step will automatically generate an "Untitled" column after data is connected to it.

Custom settings

In the left-side toolbar, first choose in the Operation dropdown menu whether our new column should show the Min or the Max.

Then choose the columns to operate on by selecting values under Range Begin and Range End.

After this, name your new column and click Show Updated Results at the bottom to save.

There is an advanced settings menu that you can expand. By default, the box is checked to Prefer numbers when running your operation. You can deselect this option, but we recommend leaving it on.

Your table will now have a column appended, displaying the Min or Max value per row, within the columns in your selected Range.

The Find min by group step calculates the minimum value in a column or a group of columns. This step is similar to the MIN function in Excel.

Input/Output

Our input data has 11 rows of order information showing us how many items were sold, the food type, the customer name, and their payment type.

If we instruct the step to find the minimum value in the Quantity Sold column, and create one new row per value in the Food Type column, we can easily find the smallest sale of each item.

Custom settings

This step has no default behavior, and must be customized to return a useful output.

First, decide whether you would like to include or exclude given columns by selecting the following columns or all columns except for.

Next, click the plus sign to choose the columns you wish to include or exclude.

If the Create a new row... option is left blank, the step will return one value per column, according to your specifications in Find the minimum...

To find minimum values for multiple attributes, you can Create a new row for each unique value in your desired columns. As before, you can include or exclude given columns, and you can click the plus sign to specify.

Unique value columns will add left to right, and sort A-Z, and your desired minimum value will display as a new column with "(min)" appended to its header.

The Find overlap step finds matching or non-matching rows between two tables.

Input/output

This step takes two tables as its input and shows you how they overlap. It will help you answer a question like, "Which rows from Table 1 do or do not exist in Table 2?"

Input 1

Table 1 shows products in our first store with three columns: "Variant ID", "Name", and "Sales Price".

Input 2

Table 2 shows products in our second store with two columns: "Variant ID" and "Name".

‍

Output

Let's say we want to find which products from Store 1 are also sold in Store 2. The step will immediately tell us this as its output. Below we can see that there are three products sold in both stores. Since "Store 1" was our primary table in the step, it's showing us the complete rows from that table.

Default settings

When you first connect your two inputs into this step, you'll see that your first input is treated as the primary table, and that rows have been matched by the lead column. By default, the step will Keep rows where any of the rules are true.

Custom settings

First, make sure the step is configured to use the correct primary table, in this case, Input 1. Then, determine how the rules should be honored. If you are setting multiple rules, the step can be structured so that any rule counts as a match or all rules must be met to be considered a match.

Now we're ready to configure the rules for how the step should identify matches. Each rule compares tables according to values in specified columns. First, select the relevant column from each table. Then, choose one of three options in the match type:

Exact match will look for an exact match, including casing. Match ignoring case will allow for differences in capitalization. Approximate match will look for a fuzzy match.

If you choose Approximate match, you can adjust the confidence percentage. The higher the percentage, the closer to an exact match this step will attempt to make. For comparison, 100% confidence is equivalent to searching for an exact match.

To add another rule, click on the + Or, another match link below the first one.

The Flip table step flips an entire table on its side, turning it 90 degrees. Imagine your column headers transform into rows. This step's functionality is similar to the TRANSPOSE function in Excel.

Input/output

The input table we'll use for this step has 3 rows and over 100 columns. Every column represents a different make, model, and year of car we have in our inventory.

By using the Flip table step, we can turn the table 90 degrees to make the previously 3 rows our columns ("Make", "Model" and "Year") and have the 100+ columns become 100+ rows. This transformed table of 3 columns with 100+ rows makes our data a lot easier to understand.

Custom settings

This step doesn't require any additional configuration. Just connect a step containing data to this one and see your table automatically "flipped" 90 degrees.

The Format dates step converts a column's date values into a new format. It's useful for standardizing date formats across various data sources' columns and making sorting by dates in ascending or descending order easier. This step can also convert between timezones and perform date rounding.

Select columns to reformat

First, select the columns that contain dates to format. Each rule card can be applied to any number of columns. The formats defined in the rule card will be used for all columns targeted by that rule.

Define the starting format

Use the first formatting menu to define the date format of the column(s) that you are targeting. Use a preset, build a custom format using the “Insert element” dropdown within the format field, or click “Auto-detect format” to use AI to find the correct starting format.

When creating a format, wrap any non-formatting character in single quotes. For example, 2024-01-01T10:30:59 should be represented as yyyy-M-d'T'H:m:s with the T character in single quotes. Punctuation does not need to be wrapped in single quotes, just letters or words that are not used to format.

If the format you have entered is valid, you will see an example date using that format below the formatting field.

Define the new format

Use the second formatting menu to define the new date format to convert to. Use a preset or build a custom format using the “Insert element” dropdown within the format field.

When creating a format, wrap any non-formatting character in single quotes. For example, 2024-01-01T10:30:59 should be represented as yyyy-M-d'T'H:m:s with the T character in single quotes. Punctuation does not need to be wrapped in single quotes, just letters or words that are not used to format.

If the format you have entered is valid, you will see an example date using that format below the formatting field.

Timezone conversion

To the left of the format fields are icons that represent timezone. The default settings will not convert timezone. Use the timezone menus to define the starting timezone and the new timezone. If a timezone token is provided in the starting format, Parabola will read the timezone directly from the input dates.

Date rounding

Use the date rounding field to round the dates during the formatting process to a specific datetime precision. Use the menu to round to the start or the end of a year, month, week, day, hour, minute, or second.

When using the week option, use the additional field to define the starting day of the week.

Rounding examples:

Round to the end of the month

• Input: February 15, 2024
• Output: February 28, 2024

Round to the start of the week. Weeks start on Tuesday

• Input: Wednesday June 5, 2024
• Output: Tuesday June 4, 2024

Round to the end of the week. Weeks start on Tuesday

• Input: Wednesday June 5, 2024
• Output: Wednesday June 5, 2024

Creating new columns

Each rule can either format the dates in their respective input columns, or create a new column for each new format to be placed in. Each column targeted by a rule can be formatting into a new separate column.

Failure settings

If an input date format cannot be formatted, use the failure settings to define what the step does.

• Fail the entire step run
• Keep the date in its original format (default)
• Convert the date to a blank
• Fallback to this value…

Columns with multiple formats

If you have a column with multiple formats in the same column, use multiple rule cards to target each format in that column. When taking this approach, be sure to keep your failure setting set to “Keep the date in its original format”

These are the docs for the older versions of the Format Dates step.

Input/output

The input data connected to this step must contain at least one column with data in a date format. In this example, we pass a column called "Sales Date" that is currently formatted as M/D/YY.

We request the New Format to be YYYY-MM-DD by selecting from the dropdown, resulting in the output data appearing as 2024-02-01.

Custom settings

When configuring the Format dates step, start by selecting the column(s) that you'd like to format. If there are multiple columns with the same starting format that you want to convert into a consistent new format, use the plus button to select multiple columns under Columns to Format. Create a rule for each unique Starting Format in your dataset and the New Format that you'd like to apply.

Note that both the Starting Format and the New Format settings offer a dropdown menu of preset formats to choose from, but you can also define a custom format by typing directly into the text box below the dropdown. Refer to the table at the bottom of this page for the supported date formatting options. Note that date formatting is case-sensitive when entering a custom format.

Specify the Starting Format of your incoming data by selecting from the dropdown, leaving the dropdown blank for Parabola to auto-detect, or entering a custom format. Parabola will only auto-detect your date column(s)' starting format if the values are uniform. Note that this feature will evaluate the date values once and set the Starting Format accordingly. It will not re-evaluate the date format every time the flow runs.

Similarly, for the New Format, you can either select from our preset format options or enter a custom format.

The Interval Start option is useful for grouping dates by a certain time series. In the screenshot below, Week is selected as our Interval Start option, which backdates each date to the start of the week (Sunday) it falls in. Choosing Day sets the timestamp to the start of the day (00:00:00), Month sets the date to the first of the month, and Year sets the date to the first of the year.

By default, the Format date step will format the dates within the existing column. If you'd like to preserve the date formatting of your existing column, check the box next to "Add New Column" and give your new column a name.

You can configure multiple date formatting rules within the same step by clicking the Add Rule link. The rules are applied top-down, so if you are formatting the same column in multiple rules, the second rule should use the New Format from the first rule as the Starting Format of the second rule.

Date formatting reference table

‍

The Format numbers step can reformat numbers across multiple columns in different formats. It can control numeric formatting options, such as, commas, currencies, rounding and percentages.

Check out this Parabola University video for a quick intro to the Format numbers step

Input/output

Let's use some of our store's sales information as the input data.

In this example, we'll want to reformat the numbers displayed in the columns "Sale Price" and "Refunded Price" to an accounting format with currency and proper decimal points.

After connecting this data into the Format numbers step, we select the column(s) we'd like to apply the formatting rules to by clicking the plus button. You can either choose to Format these columns to specify the columns you'd like to apply formatting to, or you can choose to Format all columns except to specify the columns you'd like to exclude from formatting.

Next, you can either select from the pre-set options in the dropdown or click the ... button to select from the following options:

• Numbers
• Percents
• Currency
• Accounting

By switching the default "Numbers" setting to "Accounting," our output data is updated with $ symbols, 2 decimal places, and parentheses wrapping negative numbers.

Default settings

When you first connect your data into the Format numbers step, it will auto-select the first column with numeric values that appears in the data set. It will automatically apply the "Numbers" formatting, which displays numbers  with comma groupings, 2 # Decimals, and round to the nearest. The below screenshot shows us an updated "Variant ID" column after those settings automatically formatted it upon connecting our data.

Custom settings

Upon choosing one of the default dropdown options, you can also click the ... button to set more specific formatting.

You can also add multiple formatting rules within the same step by clicking on the Add Format Rule button. Please note that your formatting rules will be applied from the top down.

Numbers

When you select to format columns as Numbers, you will have three additional settings to configure:

1. Comma grouping: You can choose whether to apply comma groupings or not. Selecting to apply comma grouping will add a comma for every thousands.
2. Decimals: You can input how many decimal places you'd like to display.
3. Rounding: You can choose whether to round to the nearest value, round down, or round up.

Percents

When you select to format columns as Percents, you will have three additional settings to configure:

1. Comma grouping: You can choose whether to apply comma groupings or not. Selecting to apply comma grouping will add a comma for every thousands.
2. Decimals: You can input how many decimal places you'd like to display.
3. Rounding: You can choose whether to round to the nearest value, round down, or round up.

Currency

When you select to format columns as Currency, you will have three additional settings to configure:

1. Currency: You can select a desired currency among the following options: Dollars ($), Euros (€), Indian Rupees (₹), Pounds (£), and Yen (¥)
2. Decimals: You can input how many decimal places you'd like to display.
3. Rounding: You can choose whether to round to the nearest value, round down, or round up.

Accounting

When you select to format columns as Accounting, you will have three additional settings to configure:

1. Currency: You can select a desired currency among the following options: Dollars ($), Euros (€), Indian Rupees (₹), Pounds (£), and Yen (¥)
2. Decimals: You can input how many decimal places you'd like to display.
3. Rounding: You can choose whether to round to the nearest value, round down, or round up.

The main difference between the Currency and Accounting options are that accounting will display negative numbers in parentheses ($5.00) while currency  displays negative numbers with a negative sign in front -$5.00.

Invert Number

The Invert number (+/-) option will switch the sign of the value, making all positive numbers negative and all negative numbers positive.

Helpful tips

• The Format numbers step can also remove formatting from numbers. For example, if you have a column that contains the value, "$4000", you can configure to format this column as a "Number" which will update the formatting of the value to "4000".

The Limit rows step filters data by keeping or removing a certain number of rows in your dataset. You can keep or remove a certain number of rows from the top, bottom, or a random position in your dataset.

Input/output

In our input data below, we have 100 rows of data.

By using the Limit rows step, we can easily filter the number of rows to keep the top 10 rows out of 100, as seen in our output data below.

Default settings

By default, this step is set to Keep the first 3 rows of data and apply that limit on published runs.

Custom settings

You have the option to switch the "Keep" selection to "Remove" instead. You may also select other options in the "first" dropdown by switching this to the "last" or a "random" set of rows. Lastly, you can change the number of rows to limit your dataset to by clicking on the "3" and typing in your desired amount.

Ignoring the limit set during published runs

Use this setting to only limit rows while you are building in a draft of your Flow. After publishing, any run (scheduled runs, manual runs, email/webhook runs) will ignore the limit imposed by this step and all rows will pass through it.

The look up rows step locates the index of the column that the value is in and appends it to a new column. This step is useful for creating connections between different data locations in a table. You can annotate where certain data is, and link it to a new column.

Input/output

Our input data is a table of daily webinar attendance for a week.

Let's try to find which day of the week is the first time we experience a maximum webinar capacity. In this case, our max webinar capacity is 25 attendees. By using the look up rows step, we can create a new column called "Max capacity hit" that shows which day of the week we first experience 25 attendees.

Custom settings

To set up this step, start by noting the range of columns it should look within. Select a column from the Range Begin dropdown and select another column from the Range End dropdown.

Next, in the field Find Value type in the one it should search for within our selected range. You can also check the box "from Column" if you'd like the step to look for a value contained in a specific column within this noted range. Checking off the box "from Column" will convert the Find Value field into a dropdown of column options.

In the New Column Name field, enter the name you'd like this new column to be called.

Lastly, checking the box for "Case" will force the matching to be case-sensitive.

The Merge columns step combines data from two or more columns into a single column separated by a specified character.

This step can serve a lot of different purposes, though it is especially useful for creating lists inside of a single cell, which can help with transferring complex data or creating lists to search against for avoiding duplicates.

Input/output

Let's say we need to combine the data from our columns "Product" and "Size" to create a new column called "Product Label" for easy label creation. The Merge columns step is how we'll achieve this.

Our input data has two separate columns "Product" and "Size".

After using the Merge columns step, we have a new column called "Combined: Orders" where  values from the columns "Product" and "Size" were merged while separated by a comma.

Custom settings

Once you've connected your flow's most recent step to this one, open up the Merge columns step window and under New Column Name, type the name you'd like your new column to be.

Then, under Merge, select two or more columns from your original data that you'd like to merge.

After that, expand Advanced Settings if you want to set a custom delimiter. Our example sets the delimiter to a comma and space  ", "  which will be placed in between the data from the "Product" column and the data from the "Size" column. If no delimiter is inputted, the step will default to merging the data in the columns with no comma or space in between.

If necessary, you can also check the box next to "Add delimiter after blank values". This is powerful if you're trying to create a JSON array or object and it's important that each object or array has the same number of commas, separating all column values including blank ones.

Common questions and how to troubleshoot

Here are a few common questions that come up around the Merge columns step and ways to troubleshoot them:

Issue: I’m seeing my rows multiplied in the results section of the Merge columns step.

Solution: It’s possible that you are matching on data that is duplicated in one or both of your sources. Check your inputs to make sure the selected matching columns have unique data in both inputs. If you don’t have a column with unique data, you may need to choose multiple columns to match with to make sure you’re not duplicating your rows.

Issue: I’m seeing a Missing columns error in my Merge columns step.

Solution: This can happen when you make changes to one of your inputs after you’ve already selected the setting in your Merge columns step. You’ll need to either reselect your matching columns in that step or pull in a new Combine tables step.

The Merge duplicates step allows you to group rows in one or more columns and merge their values.

Input/output

The data we'll input into this step has an identifier type to deduplicate and its corresponding columns of data.

The step provides us with output of deduplicated data (based on a unique identifier value) with the merged values separated by a Delimiter in their corresponding columns.

Default settings

By default, Parabola will map your first column to the unique value selector, and your second column to your Merge selector.

The Delimiter will be automatically set to a comma (,).

Lastly, the option to Ignore blanks will be auto-selected on.

Custom settings

First, decide whether you want to include or exclude columns, by selecting these columns or all columns except.

Next, click the plus sign to select the columns you would like to merge.

Then, use the Per unique value in selector to set the column which will act as your unique identifier.

You can choose whatever Delimiter you want. By default, it is a comma (,) but you can choose whatever works best with your data, such as a space ( ), a dash (-), or more.

Selecting the option to Only merge unique values keeps all values unique. For example, differently formatted phone numbers will not roll up into one merged number.

Ignore blanks is selected by default and treats blank cells as if they are not data points. Unchecking this option includes blank cells into your merged values, separated by your Delimiter.

The Pivot columns step takes a column and turns its values into multiple columns. It is particularly useful when needing to turn order line item information into a unique row per order.

Input/Output

In the example below, our input data has duplicate values in the "Order ID" column, plus other columns specifying "Variant Option" and  "Variant Value".

We'll want to create a unique row per "Order ID" where the "Variant Option" becomes the column headers and "Variant Value" becomes the cell value in the corresponding columns. The Pivot columns step accomplishes this for us.

Custom settings

Once you connect data to the Pivot columns step, you'll select a Column to pivot into new column headers, and you'll select a Column with values to fill in

In the example below, we selected the "Variant Options" column to be the pivot column, and we selected "Variant Value" column to be the value column. If no "Variant Value" exists for the corresponding "Variant Options" column header, the cell is left blank.

‍

The Remove duplicates step removes rows with a duplicate value in any column you choose.

Watch this Parabola University video to see the Remove duplicates step in action.

Input/output

Our input data has 100 rows with these 4 columns: "Webinar ID", "Registrant name", "Registration date" and "Registration time".

After using the Remove duplicates step to remove duplicate rows in the "Webinar ID" column, we went from 100 rows to 87. This tells us that we have 41 distinct Webinar IDs in our table. You can see from the screenshot above that some Webinar IDs like 532444816-8, 061561006-4, 966924489-7, 905788038-5 are repeated at least twice. It's those duplicates that this step helps remove instantly.

Custom settings

When you connect data into the Remove duplicates step, you'll see a notification at the top of the step configuration that says, Per unique value in these columns: select from menu.

First, select which column to look for duplicates. Clicking into the select from menu dropdown will expand a columns list to choose from.

You can also adjust the number of duplicates you'd like this step to keep. By default, this step will only keep 1 row per unique value, though can be changed to whichever number you'd like.

A common task is to take a list of data and keep the first 1, 2, 3 or N number of entries depending on a certain order. For example, say we wanted to take our list of webinar registrants and keep the first 2 registrants per webinar to recognize the customers who registered early. First, we'll make sure that the input data is sorted by "Registration time" where the oldest times display at the top. We would then update the Keep the first _ to 2 and you'll see, as opposed to the example above, it keeps 2 rows per Webinar ID.

Helpful tips

• If you want to dedupe based on data existing across two or more columns, use a Combine columns step before the Remove duplicates step. This creates a column that has the values across your columns. You can then select that newly created column in your Remove duplicates step.

This step is now part of the Edit columns step. If you are using this step in an older Flow, it will continue to work but will not be available for new Flows.

Input/output

Our example input data has orders data that informs us of the units sold per product variant. We have a column called "ID", but we'd need this to be more exact about what "ID" this actually is for our colleagues.

We can use the Rename columns step to give the "ID" column a more descriptive column name, such as "Variant ID".

Custom settings

The configuration for this step is thankfully very simple. Start out by selecting the column you'd like to rename from the dropdown link select from menu and provide a new name in the type value... field. Select Show Updated Results at the bottom of the step settings, and you're all set to go!

Helpful tips

• You can add as many rename rules as you need by clicking on the Rename Another Column link below the step's first one.
• If you have to rename a lot of columns (10+) or the columns have ambiguous names that are difficult to locate in the select from menu dropdown, you could create a CSV file with the desired names as headers (with no row data). Then, you can use the Stack tables step to combine this headers table at the top with the rest of your main data underneath.

‍

This step is now part of the Edit columns step. If you are using this step in an older Flow, it will continue to work but will not be available for new Flows.

Input/output

In this example, the input data we'll use has orders information that displays the units sold per product variant. However, our columns are in a randomized order that's difficult to understand quickly.

We can use the Reorder columns step to move our columns into whichever form that makes the most sense to us. As seen in the below screenshot, we've grouped our product data together with the variant data, and then placed the "Units sold" column at the very end (right-side).

Custom settings

When you first connect your data into the Reorder columns step, it'll display your columns in their original order.

You can click down on a specific column and drag it into the desired placement. You may also type in a number representing the position of the column. The placement of the other columns will automatically update based on your input.

The Replace with Regex step matches patterns using Regular Expression to find or replace values. Regular Expressions, or RegEx for short, are useful for matching patterns in text or numbers (or anything, really). RegExr.com is an excellent resource to use when composing your regular expression.

‍

Before you jump in...  We recommend exploring the "Extract text from column," "Find and replace," and "Clean data" steps. These steps are often able to accomplish the same result without writing a Regular Expression. 

Input/output

For our input data, we'll use a list of nine Webinar IDs. The number displayed after the hyphen "-" is the number of attendees that can sign up for the Webinar. We're looking to extract the Webinar ID and display it in a new column.

Our output after using this Replace with Regex step is a new column "Remove Attendee Count" with the numbers that display before the hyphen in the original Webinar ID values.

Custom settings

To start configuring your first rule, you'll select the column we should apply the RegEx too. You can also select to search through All from the Column dropdown.

Then, input the expression we should look for in the Expression field.

If you'd like to replace the value with something specific, put that value in the Replace Value field. In my case, since I wanted to remove the found expression and only retain the remaining number, I left this field blank.

You can click the checkbox to "Add New Column" if you want to preserve the original data column but display the extracted value in a new column. In my example, I selected to "Add New Column" and placed my column name, "Attendee count" in the New Column Name field.

You can create as many RegEx rules as you'd like in a single Replace with Regex step. To do so, click the button to "+ add rule".

Helpful tips

Again, we recommend RegExr.com as a useful tool when working with RegEx. We particularly find their "Community Patterns" section useful where you can find RegEx patterns that others have used before.

You can also consider experimenting with AutoRegex.xyz, which is a useful app that uses GPT-3 to convert plain English to RegEx.

Characters

• . any character, except a newline
• \w any word
• \d any digit
• \s any whitespace
• \W anything except a word
• \D anything except a digit
• \S anything except whitespace
• [abc] any of a, b, and/or c - you can use dashes in here too such as [a-z] or [0-9]
• [^abc] not a, b, nor c - you can use dashes in here too such as [a-z] or [0-9]
• [a-g] any character between a and g
• [0-5] any digit between 0 and 5‍

Quantifiers and Alternators

• a* any amount of the letter a in a row. 0 or more
• a+ 1 or more of the letter a in a row
• a? 0 or 1 of the letter a
• a{5} exactly 5 of the letter a in a row
• a{2,} 2 or more of the letter a in a row
• a{1,3} between 1 and 3 (inclusive) of the letter a in a row
• a+? 1 or more of the letter a in a row, but match as few as possible
• a{2,}? 2 or more of the letter a in a row, but match as few as possible
• ab|cd match either ab or cd in a cell

Anchors

Anchors help you define how an expression is related to the beginning or end of a cell

• ^abc the cell that starts with abc
• def$ the cell that ends with def
• ^$ a blank cell

Escaped characters

Using a backslash, you can indicate invisible characters, or escape any character that normally has a special purpose

• \. escape a dot so that it is seen as an actual dot
• \* escape an asterisk so that it is seen as an actual asterisk
• \\ escape a backslash so that it is seen as an actual backslash
• \t find a tab in your text
• \r find a newline in your text
• \n find a different type of newline in your text

Groups & Lookarounds

Groups are used to capture bits of text and then interact with them in the replacement function.

• (abc) capture the group that contains abc within a cell
• $1 reference the first capture group (in the "replace" field). Use $2 for the second capture group, etc.
• $& reference all capture groups (in the "replace" field)
• ‍(?:abc) non-capturing group that contains abc within a cell‍
• (?=abc) positive lookahead‍
• (?!abc) negative lookahead

This step is now part of the Edit columns step. If you are using this step in an older Flow, it will continue to work but will not be available for new Flows.

Input/output

In our input data below, there are four columns: "Variant ID", "Product Name", "Variant Name", and "Sale Price".

After connecting our data into this step, it produces an output of the two columns selected to keep: "Variant ID" and "Product Name".

Default settings

When you first connect your data into the Select columns step, it will auto-select to Keep instead of Remove columns. It will also default to keeping your first two left-side appearing columns in a table.

Custom settings

To customize these settings, first decide whether you'd like to keep or Remove columns. We generally recommend selecting the option with fewer columns to select, since this will save you time when finalizing the step's settings.

‍

After this, you'll then customize the column selection. Clicking on the blue dropdown box brings up a list of columns. You can select columns directly from that dropdown, or type the name in directly. If a column is selected in the dropdown, you can hit "return" or "enter" on your keyboard to select it.

To remove columns from your selection, click the "x" next to the column name in the blue box or use the backspace key on your keyboard near it.

Helpful tips

• Depending on your data source, usually when working with APIs, the data source's column structure varies. If a column you selected to remove or keep no longer exists in the input, the step will error. Try to Keep or Remove selections in a way that'll focus on the most consistent columns. We generally find that keeping columns are a safer bet. If you hit this error when running your flow, you can use an Insert static text columns step with its advanced setting enabled (to only add columns if it does not exist in your input).

The Sort rows step sorts the entire table in ascending or descending order based on values in the column(s) you choose.

Input/output

The example data we'll connect to this step has a table of 100 webinar registrants.

Let's say we're wanting to view this data with the top being the newest registrants by registration date. We can do this by setting a sorting rule where the values in the column "Registration date" are sorted in Z-A (descending) order.

Custom settings

When you connect data to a Sort rows step, you'll first select a column with the values you're wanting to sort the data by.

You can then choose whether it should be sorted in A-Z (ascending) or Z-A (descending) order.

If you'd like Parabola to ignore blank values during sorting, you can check the box to "Ignore Blanks". Without checking this box, blank values will be considered the lowest value. Checking this box may be useful when sorting your data in ascending order.

If you require multiple sorting rules, click on the  Add Sort button to add multiple rules in. The rules are applied top-down, so the first rule will be the first sort, and the second rule will sort entries within each group that has identical values in the column used in the first sort.

In the example below, we're first sorting by "Registration date" in descending order so we can quickly see the most recent registrants at the top. Then, we sort by "Registration time" in ascending order. We'll see that for rows where "Registration time" is 3/20/2024, the "Registration time" is properly sorted in ascending order.

Helpful tips

• A key concept to remember when sorting data in Parabola is that you are sorting rows, and not cells.
• Ambiguous dates (such as MM/YYYY) are difficult to sort. Try using a Format Dates step to convert your dates into a less ambiguous format before sorting.

The Split column step splits a column into one or more  based on a specified delimiter, such as a comma.

Input/output

Our input data below is a single column called "Data".

We'll use the Split column step to split this single column into three different columns using the delimiter of a dash mark "-". The output data will look like this:

Custom settings

First, in the dropdown select from menu, choose which column to split into one or more.

Next, decide if you'd like your data split into new columns or  rows by clicking into the linked dropdown menu columns.

You can separate a column's value based on any delimiter by typing into the section type value... whether it be a comma, semi-colon, dash, space, number, letter, or special character. In our screenshot examples, the "Data" column is split into three columns based on a dash mark -.

Helpful tips

• If you'd like to join a dataset using this step to another one based on the step's newly-generated column output and choose to separate values by a comma, be sure to include the space after the comma in order for values to start with the letter or number. For example, splitting apart a column of orders like "shirt, pants, hat" by the comma without a space — like (,) versus (, ) — will create the output of rows "shirt", " pants", and " hat".

Use the Split names step to split a full name value into smaller parts (first, middle, last, etc.).

Input/output

Your input data should consist of a column of full names.

Your output will have multiple new columns of your names, separated into their parts, depending on the parts of the name you chose to split.

Custom settings

First, select the column that has the names you'd like to split by selecting it in the Column to Parse dropdown.

Next, pick the category that you want your full names broken down into by selecting them in the Choose Parts of Name dropdown.

You can select multiple parts in this dropdown, including Salutation, First Name, Middle Name, Last Name, Suffix, and Full Name.

If you select a part that does not exist in a name, the row's column will appear blank.

Helpful tips

• Names ending in a comma may not be parsed.
• Names such as "Mister" may be parsed as the salutation, as opposed to the name.

The Stack tables step provides the ability to vertically stack any table(s) on top of one other.

Input/Output

The Stack tables step needs at least two data inputs. In the example below, we have two tables that both have the columns "Product Name" and "Food Type".

After using the Stack tables step, these two tables are now stacked one on top of the other.

Custom settings

First, connect at least two datasets (or flow branches) to this step on your canvas.

Then, double-click on the Stack tables step to open up a pop-up window. This displays your datasets combined where the first connected dataset is on top while the second  dataset is right below.

By default we stack the tables vertically based on the order of columns. If you prefer, you can click a checkbox to have Parabola match up data by column names. If your inputs are integration steps or APIs, we recommend selecting this option to match data by column names for a more stable flow, in case the API/integration decides to change up its column ordering.

Helpful tips

• This step requires at least two inputs. Make sure you have at least two arrows connected to this step. You can have as many inputs as you need. There's no maximum number of inputs.
• The order of the arrows matter. This is how we know what order to stack the tables in. The column headers from the first input are the headers used by the step's output result. The numbers you see in the arrow indicates the order. Your first input is the one with a blank arrow.

• If you have fewer columns in your first data input than you do in your other ones, the step's result won't retain those extra columns from the other data inputs.

The Standardize with AI transform step evaluates data and standardizes it based on examples that you predefine.

At launch, you can use Parabola AI steps at no extra charge to your team. After a beta period, we’ll move to a usage-based model for Parabola AI. (You will not be charged retroactively for usage during the beta period.)

Examples of categorizing data with AI

• Clean a table where the names of columns are not always consistent, such as packing lists from multiple vendors. Convert columns with names like “Product ID,” “sku,” and “Style” into “SKU”
• Take a list of clothing items with inconsistent size names and standardize those names. Convert sizes with names like “Med,” “M,” and “mdium” into “Medium”
• Correct words that are misspelled, contain errant spaces, or have undesirable casing.

After running the step and seeing how the AI responds, it’s normal to modify your examples and re-run it to fine-tune your results.

How to use this step

Selecting what to evaluate

Start by choosing whether you want to clean up the names of your columns, or the values in a specific column.

1. Standardize column names: the AI will rename columns based on examples that you provide.
2. Standardize values in a column: The AI will clean values in a specific column using examples that you provide as a guide. The cleaned values will be displayed in a new column, next to the column that the AI evaluated

Setting up examples

This step is designed to use examples to guide the AI to find values or column names to update. The better the examples, the better the AI will be at evaluating and standardizing your data. When evaluating a value, the AI will use the examples, as well as the desired new value, to make decisions. It will also infer and guess at names that you may not have given an example for.

Include as many examples in each value as you would like, and separate them with commas. Each “Value” in the step settings should correspond to a single new value that you want to use as a column name or a cell value.

Fine tuning

Open the "Fine tuning" drawer to see extra configuration options. Using this field, you can provide additional context or explanation to help the AI deliver the result you want.

If the AI is not quite catching everything you want, try using this field to explain how to find those edge cases. You can explain things in the same way that you’d explain them to another person.

Helpful tips

• Sometimes you’ll see a response or error instead of a result. Those responses are often generated by the AI, and can help you modify the prompt to get what you need.
• Adding more examples in each value can help guide the AI

The Sum by group step calculates the sum of all values in one or more columns. This functionality is similar to the SUMIF function in Excel.

Take a look at this Parabola University video to watch the Sum by group step at work.

Input/output

Our input data has 39 rows of order information showing the quantity sold, food type, customer username, and payment type per order.

After we connect this data to the step, it'll produce an output of 9 rows that sum quantities sold per food type in the "Quantity Sold" column.

Default settings

Once connected to your flow, this step will auto-select a column to find the total amount per data group/category like quantity sold per food type. The result will display your grouped values and the total sum of them as a new table.

Custom settings

If you'd like to choose other columns for your sum calculation and which rows the calculation is grouped by, you can change the auto-selected ones by clicking into the blue boxes on the left-side toolbar.

You can also select multiple columns to sum and multiple rows to group by, or choose which columns to exclude.

The Unpivot columns step will take every column that you choose and line them up into a single column called "Value". A new column will be made next to it called "Type" that holds the value of the column header that  data points used to be in.

Input/output

The input data we'll use for this step is a table of three columns: "Store Location", "Profit", and "Loss".

Below is a screenshot of our output data after using this Unpivot columns step. It has taken our columns "Profit" and "Loss"  and merged them into a single column called "Value". The "Type" column holds the value of the column header these values came from.

Custom settings

After you connect your data  into this step, select the dropdown menu of Unique Identifier Column to choose the column (type of data) you'd like to use as an identifier for this data set. This should be the column you don't want pivoted, and will be used as a key if the data needs to be pivoted back into shape. This could be a customer ID, store name, store location, customer full name, or more.

Then, select the remaining column headers (data categories) that you'd like to merge the values of those columns into a single column.

Finish by clicking Show Updated Results to save the step's settings.

Helpful tips

• Optionally, in the Pivot These Columns section, you can choose to select the columns to exclude from the unpivot operation.

‍