Pair Eyewear saves $50k annually by automating their inbound PO process

In the growing world of direct-to-consumer eyewear, Pair Eyewear has made a name for itself by giving consumers what they’re looking for: optionality.

Not only are their frames customizable, they’re also priced so that you never have to get stuck in an eyewear rut: Instead of relying on one pair of glasses, Pair Eyewear makes it possible to have a pair for a special night out, your favorite season, and for everyday use. They’ve even collaborated with major museums and beloved franchises to bring frames with tons of personality to market. 

But with the success they’ve found since launching in 2019, they’ve felt some natural growing pains — this is something that became particularly apparent when they expanded their wholesale business and were receiving tons of inbound PO PDFs via email.

Between hard-to-access data and PDFs clogging their inboxes, Zach Wilner, who leads data and analytics at Pair, and Zach Headapohl, the Senior Manager of CX Operations, were in the market for a solution.

The challenge

Wilner on the data side remembers the early days of their growth all too well. “We were using Narvar as our returns management system,” he recalls, and they needed to be able to pull data from that system for internal reporting. But there was a major limitation: They could only export data as a CSV or send to an email address.

This seemingly small issue created a significant bottleneck in their data pipeline, preventing the team from accessing crucial information about orders in real time.

Meanwhile Headapohl on the ops side was grappling with a different set of problems. As his team expanded, so did their need for analytical firepower. “We needed a sandbox,” Headapohl explains, “somewhere we could prototype and iterate without constantly competing for engineering resources.” The operations team found themselves in a constant tug-of-war, trying to balance their growing needs with the limited availability of technical team members.

And as Pair expanded their wholesale business, they faced a new hurdle: an influx of purchase orders arriving as messy, hard-to-work-with PDFs. The volume was so high that they were considering hiring a full-time employee just to manage the process of getting that data cleaned and sent to their 3PLs. 

The solution

When Wilner first discovered Parabola, it was through Narvar, their order management system. They were unable to get the data from that platform into Snowflake, so they were looking for something that could push data from email to S3 (they had a home-grown ingestion system from S3 into Snowflake). 

But what started as a solution to one specific problem soon blossomed into something that has become a sort of foundational tool for the ops team. Headapohl and his team quickly realized Parabola’s potential beyond simple integrations. It became their sandbox — a place where they could rapidly prototype and build automations to support their processes and data needs without relying on engineering resources.

It essentially created an army of citizen developers out of mostly non-technical ops folks.

For Headapohl, the full scale of Parabola’s power became apparent as he worked through a hefty project in which he was trying to merge returns data with quality data: “So all of this would be pretty easy to do in Excel if I had a super computer…but I’m dealing with like a half a million records across 36 different fields in some cases, and am standardizing them so I can stack them and then create unique identifiers.” 

“To have something to power that is pretty rare,” Headapohl explains. 

Most recently, they’ve been using Parabola’s PDF parsing capabilities to solve Pair’s wholesale order processing challenge. 

Typically, Pair receives PDF POs from retailers via email. They had to manually convert them to .csv so they could 1) send it to the 3PL and place the order for shipment and 2) let their planning team know how much inventory is going out the door so they can track revenue metrics. 

Because their wholesaler doesn’t have EDI or any electronic transfer capabilities, they were stuck in a very manual process.

This led them to scope Parabola’s PDF parsing step (which combines OCR vision technology with AI to read and contextualize PDFs with extreme accuracy) to see if it could start doing this work for them. 

Spoiler alert: It could. 

Here’s how they’ve now automated this whole process — instead of hiring someone to manage it full time:

1. Retailers send POs to a Pair email
2. A Pair team member then reviews the PDF before sending it through to Parabola for upload
3. Parabola scrapes the email for a PDF, then automatically cleans it and puts it into a spreadsheet format
4. From there, it lands in S3, and from S3 they pick it up and bring it into Snowflake
5. Teams at Pair then access this via looker and are able to action on the PO

“It’s been very, very accurate from day one. I actually barely put any prompts into the AI — I just kind of turned it on, added the columns I wanted, and set it free. It’s a really big unlock for us,” Wilner said. “These PDFs can be six pages of really poorly formatted order data that can be hard for even a human to read.”

The results

As Pair Eyewear continues to grow and evolve, Parabola has become an integral part of their technology stack. 

“We can’t ever rip this out, the ops team has become so dependent on it,” Wilner says. This only seems to improve his working relationship with Headapohl and ops as the person leading data at Pair. 

Pair’s experience with Parabola started with a simple Flow to solve a critical (but simple) problem they were facing. Now? The whole ops team is enabled with a data tool that gives them the power to automate their workflows, clean and access their data, and iterate on team-wide processes without the support of engineering or data teams. 

Plus, Pair can report: 

1. At least $50k annual savings because they didn’t need to bring on full-time resources to manage wholesale PO data management.
2. Save 20 hours/month on reporting. As Headapohl puts it, "Parabola allows my team to process large, multi-source reports saving hours each month by avoiding the laborious steps required to replicate similar transformations in Excel.”
3. A paper trail that documents five critical internal processes. “You end up with step-by-step instructions on how to run some of the task work, which ultimately reduces the time and effort to accomplish that task work,” Wilner says.

And the benefits don’t always have to come from the most complex use cases according to Headapohl : “some of the stuff we do with Parabola is really simple, but tedious  to execute manually in Excel. I think we’ve all asked, what is the one thing that is keeping this data from pivoting or merging the way that I need?”

To bring us home, Wilner told me about someone unfamiliar with Parabola that they brought into the PDF project and immediately threw into the weeds. His takeaway?

“This tool rocks.”