Rewinding the clock back over a decade to their time at Columbia University, it seems obvious now that Shensi Ding and Gil Feig were destined to become co-founders. Meeting as freshmen in the university’s engineering program, the two quickly struck up a close friendship. Senior year, they even teamed up as the Engineering Student Council’s President and Vice President — a sneak peek at what was to come later.
But as “Pomp and Circumstance” played and they tossed their graduation caps in the air, the two went their separate ways — Ding for the investment banking path at Credit Suisse, and Feig off to build software at LinkedIn.
Over the years, they kept in touch and started grabbing weekly dinners together at Sweetgreen when they both lived in San Francisco, working for tech startups. One dinner, in particular, would crack open a door to the founder’s path.
“Frankly, Gil looked terrible. He was working on a particularly miserable integration project that was killing him at work — and he was listing off all the same headaches that I’d heard the engineers at our company grumble about, too,” says Ding.
At the time, the two worked for startups in very different industries. Ding was Chief of Staff at Expanse, a cybersecurity company that would track IP assets across the internet (it would later be acquired by ~$1 billion by Palo Alto Networks). Meanwhile, Feig was the founding engineer and Head of Engineering at Untapped (called Jumpstart at the time), a recruiting tech startup. While Expanse was trying to integrate with ticketing systems like Jira and ServiceNow, Untapped was integrating with fragmented applicant tracking systems (ATS).
The two were doing what old friends do over dinner — catching up about work. But what they didn’t realize at the time was that they had stumbled upon one of the biggest pain points across B2B companies of all sorts of industries. Put simply: integrations are a necessary part of building software today. But building, maintaining and scaling these integrations just plain sucks.
Cut to today, and the accolades for their company Merge have rolled in. It’s the G2 Leader for Unified APIs. It nabbed a slot on Forbes’ Next Billion-Dollar Startups list and both co-founders were named to the 30 under 30 list. Now a Series B startup that’s raised $75 million so far, with 7,000+ customers including the likes of Calendly, Gong, TripActions and AngelList, the upward trajectory in just three years has been the stuff startup dreams are made of.
But how did the duo actually build an extra-strength Tylenol for one of the biggest headaches in B2B SaaS? Let’s return to that fateful dinner.
EXCAVATING THE PROBLEM
Ding further unpacks the throughline that her company, Expanse, and Feig’s company, Untapped, were both struggling with. “Expanse’s product would surface potential security risks as issues. The problem was that people wanted to be able to export those issues into whatever ticketing system they used — but we didn’t have any integrations. Even though we had the better product, we would lose to a competitor with a subpar product because they had more robust integrations,” she says.
Plan A seemed simple enough — rather than try to build these internally, Expanse initially outsourced to a consultant. “But we couldn’t use any of the code, and had to just start all over again.”
Onto Plan B — Expanse hired a team of full-time engineers, tasked to exclusively focus on building integrations. “But it wasn’t just a matter of checking an integration off the list. Once you build your first one, you have to keep maintaining it. Because every single time a new customer links the integration, there’s a possibility that there will be edge cases. Now we have to have a never-ending amount of customer success and engineering resources just to maintain these integrations,” says Ding.
Meanwhile, over at Untapped, Ding recalls that Feig was also banging his head against the wall. “When Untapped built their first integrations, what they found was that there were a lot of end-user errors. Someone would enter the API key incorrectly, or give you an API key with incorrect permissions. All of this would look like a bug on the Untapped engineer’s side — but it wasn’t. It’s exceptionally hard for customer support to go to your customer, who’s maybe paying six or seven figures for your product, and say, ‘Hey, you made a mistake,’ without involving engineers so you can be 1000% confident that it wasn’t your own error,” explains Ding.
Different companies, different industries, but mirror images: “It was just a fundamentally painful problem, not only for the sales side, but also for the engineering side. If you don’t build these integrations, you can’t sell your product. But building and maintaining these integrations is so time-consuming that it prevents engineers from even working on your core product,” says Ding.
The fact that Ding and Feig were approaching the same problem from different angles turned out to be a major competitive advantage. “I was fortunate that Gil was running into this same exact problem but in a very different category of software. It allowed us to think about solving this problem from an entire market perspective, rather than an industry-specific perspective," she says.
We didn’t have a solution at first — we just knew that this was a big problem, and the solutions out in the market weren’t solving it.
The duo continued to talk about the problem over dinner for the next few months. Their vision started to crystallize: Build something that every company would want to outsource their integrations to.
VALIDATING THE IDEA
The eventual co-founders had stumbled upon what they believed to be a huge problem space — but they needed more ammunition than their own experiences with thorny integrations.
So over the course of six months, they sent hundreds of cold emails and ended up talking to 100 different companies, sitting down with folks in every function that dealt with integrations. This wasn’t to pitch any sort of solution — Ding and Feig didn’t have one yet. “We wanted to see how many types of companies this impacts and whether there could be a uniform solution to solve this for everyone,” says Ding.
While they talked with as many companies as would agree to a meeting, it became clear that the integrations problem most acutely affected B2B companies. “People were paying a lot of money for vendors and were deeply unhappy with the quality. It was also a problem that would continue to spiral — the number of B2B apps keeps increasing, and there’s going to be more and more fragmentation across the market,” says Ding.
Just by looking at how the market was moving, we knew someone needed to solve this problem. We wanted to make sure it was us.
Talking to folks across the org proved to be a worthwhile use of time — there were plenty of folks outside of engineering who were experiencing integration headaches. It was a full company problem:
- Sales teams struggle when deals are blocked because of a lack of integrations
- Partnership teams spend months trying to establish relationships with API providers in order to get access
- Product management and design have to research and plan the integrations, prioritizing one customer’s vendor over another
- Engineering has to build and maintain the integration, and help investigate whenever customers run into issues
- Customer success spends hours onboarding and supporting customers with integrations, and has to maintain documentation
After each conversation, the founders took a bunch of notes in Google Docs and identified the central theme of the conversation. The Merge founders knew they were onto something when each post-sales member they talked to parroted back similar themes: “They would say, ‘Even when the integration is live, I still have no idea what’s going on with it. Every day I have to get an engineer involved,’” says Ding.
Most companies the founders talked to hacked together some sort of inelegant workaround. “A lot of them would use tools like DataDog to monitor the integration if something went wrong or had failed. But this couldn’t be managed by a customer success person — it had to be an engineer. And that engineer would then go look at Sentry to see what part of the code had been impacted, and then resolve it and go back to customer success again. But there was no enablement for customer success to figure out the problem on their own,” she says.
FINDING A 10X SOLUTION
Six months in and 100 customer conversations later, Ding and Feig still weren’t quitting their day jobs. “We could have quit our jobs sooner. By the time we talked to 100 companies we had high conviction in the problem, but we were scared to leave our previous companies,” recalls Ding.
A lot of people just start a company to start a company. It took us a long time to decide to do it.
They had high conviction, but what they didn’t have was a solution — yet. “We talked about a few different ideas for what would solve this problem, but what we kept coming up with still put the onus of building the integration would still be on the company. Why would anyone buy that?”
So how did they finally land on the unified API solution? While Ding would like to tell a magical story about the aha moment, it was much less cinematic. “I think we were just throwing out a bunch of different ideas. And one of us mentioned unified API, and suddenly it started to click,” she says.
Why? They were finally circling a 10X improvement, not a fractional one. “Because it’s one integration, and your engineers don’t have to do the work of building and maintaining dozens or hundreds of individual integrations. With the Merge dashboard, your customer support team can also have full visibility into everything without getting your engineering team involved,” she says.
With Merge’s unified API, they could provide access to the APIs of multiple software providers (like CRM or accounting), providing an abstraction layer that simplifies authentication, data transformation and data syncing.
With their solution in mind, the founders circled back to all the companies in their Rolodex who had sat down in the initial idea validation phase. “We found that nine out of ten companies were like, ‘If this works (and it’s going to be hard), absolutely we will use this product,’” recalls Ding.
Next, they started talking dollars.
We asked for a price point they’d be willing to pay. They didn’t say $10k. It was $50k, $100k — even $300k. Hearing their responses was when we knew there was a huge untapped market.
Here’s how the back-of-the-envelope math worked: “You’d talk to a VP of Engineering and they’d explain that they are paying three or four engineers just to build integrations, and you could re-allocate them to much more important projects. You can also add in a percentage of closed-lost deals that are now viable because of new integration offerings,” she says.
With that vote of confidence, it was time for the founders to quit their jobs and start going fully heads-down into building. The timing couldn’t have been worse — it was June 2020, at the height of the COVID-19 pandemic. “I don’t recommend this for other founders, but I gave my notice at Expanse and told Gil, ‘Hey I quit, so you’re going to have to quit too.’ He was like, ‘Dude, why did you do that?’” laughs Ding.
But the momentum was too great to ignore. “The timing felt like if we missed this window, it was going to be too late. We had to take our shot.”
BUILDING FOR A MASSIVE TAM
“We just started working in Gil’s apartment, coding all day,” says Ding. But her computer science degree had gathered dust over the years. “I hadn’t coded in 5 years, so Gil sat next to me for three weeks and re-taught me how to code. I didn’t want to just farm the work off to him, I wanted to help build the product as a full-stack developer,” she says.
And the duo made a bold move from Day 1: they were going to build for multiple categories at once, rather than start narrow and then grow their reach. “This was a very intentional choice to signal to the market from the start that Merge is multi-category. We didn’t want to get siloed into one category,” says Ding.
The team spent almost a year in stealth building the product. “It was important to us to launch with a high-quality product – it’s very easy to lose customer trust here, and we weren’t only going to impact our own customers, we were going to impact our customers’ customers.”
It was a risky gamble at the time to build heads-down for a year for two categories without obvious customer overlap, versus shrinking the scope to gain confidence that they were on the right path. To start, they focused on integrations for HRIS (human resource information system) and ATS (applicant tracking system) tools — a very intentional choice at the time. Here’s why: “We chose these two because the kinds of problems that we would have to solve were very diverse. They would require a different sales motion, and we wanted to experiment with different price points,” says Ding.
- ATS tools (like Greenhouse, JazzHR and Freshteam) are deep, complex integrations that are hard to support. There’s a ton of data — one ATS instance for a midsize company could have millions of candidates stored in it. “The volume that we were dealing with would allow us to build for scale. Could our databases handle that load? There’s also more customizability in ATS — we knew that if we eventually wanted to add support for more categories, like CRM, we needed to be ready for that, too,” explains Ding.
- HRIS tools (like Rippling, Gusto and Workday) have a broader use case that’s fragmented — and there’s less data volume. “But there’s a wider pool of companies that you can sell an HRIS integration to,” she says.
A few months into building, in August 2020, the founders raised a $4.5 million seed round, led by NEA with angel investors including MuleSoft and Cloudflare’s CEOs, plus the founders of their former companies, Expanse and Untapped. “We knew it was time to raise capital when Gil and I realized there was so much work to do and we needed to multiply our efforts,” says Ding. “We spent a lot of time building so that investors could see the demo, but we realized that it was more compelling to tell the story of where the product could go over time.”
ACQUIRING INBOUND & OUTBOUND CUSTOMERS
Fresh from a year of heads-down building, the founders had managed to nab their first customer: Drata, a cloud compliance software company. The early investment in quality paid off. “They used another Unified API solution at the time, but they moved over to us because our product and customer support were much higher quality, and we adopted a very proactive approach to customer support,” says Ding.
Even after months of careful coding, there was still a huge learning curve when it came to building for a real-life customer, she recalls.
That first customer is like a tough parental figure who’s just disappointed in you all of the time.
The Merge founders didn’t know it at the time, but this particular first customer turned out to be exactly what the startup needed to push harder and faster. “Drata became a billion-dollar company in one year. They needed a level of speed and velocity of execution that pushed us a lot,” says Ding.
Finally, Merge emerged from stealth in April 2021, with a self-serve motion and a free tier to let the market come to them. “We started getting inbound interest from companies like AngelList. We got 1,000 signups in the first 3-4 months through a mix of SEO and some content marketing,” says Ding. Product-market fit was clearly well within their sights.
Getting so many customers upfront was extremely overwhelming. You’re dealing with so many bugs that pop up all at once — but once you get rid of those bugs, they’re gone.
Even though the Merge team was a big believer that inbound would be a powerful channel, the founders still dedicated immense amounts of time to outbound selling. In fact, they did all the sales conversations themselves for the first eight months before bringing on a head of sales. “Start with a lot of outbound — there’s no replacement for that learning. You just have to reach out to as many people as possible,” says Ding.
Ding and Feig threw themselves into their new sales roles. “For probably 4-5 months, from 10 am to when we left the office late at night, all we did was sales meetings, booked back-to-back all over the world. Founder-led selling creates a tighter feedback loop that makes your product better,” she says.
When you do founder-led sales you’re not just learning how to sell your product, you’re learning how to tell a story. This is essential to sell the vision to investors or to candidates you’re trying to hire.
But these back-to-back meetings were no lather-rinse-repeat operation. “We were A/B testing every single meeting. How can we tweak the messaging, the demo, or the persona we meet with, or what price point makes sense? And after every single meeting we would quickly reflect on what worked, what didn’t and how we could improve the next time,” says Ding.
The goal was to amass a ton of learnings that they could eventually pass over to a head of sales hire. Here’s her summary of what worked and what didn’t:
- What worked: Demoing. “What always pissed me off as an engineer was going into a vendor meeting and not being able to see the product. Once prospects could see how intuitive, yet powerful, our product was, they would be much more interested in using it themselves.”
- What worked: Self-serve. Along similar lines, getting customer’s hands on the product sooner, rather than later, was key. “There was no barrier to entry to try out the product and validate that yes, Merge works. And then we can come in for a sales conversation to upgrade.”
- What didn’t work: Meeting with the user. “We found that talking to certain personas didn’t get us closer to a sale. Meeting with an IC engineer or post-sales folks didn’t work — we needed to meet with someone with greater buying power, like a head of product or a head of engineering.”
A year after raising its seed round, Merge raised its $15 million Series A, and announced partnerships with BambooHR and Lever. The product hailed 50 integrations and 600 customers at the time. And just about another year after that, Merge raised its $55 million Series B in fall 2022, in the midst of a much trickier fundraising environment. It was hard to argue with the trajectory the company is on — growing ARR 30X over 12 months and amassing 7,000 customers. Along the way, the company expanded its integration support to include not just HRIS and ATS, but also accounting, CRMs, project management, ticketing, file storage, and marketing automation ticketing systems.
And just a few months ago, the company announced the launch of Blueprint, an AI-powered tool that speeds up the development process for Unified APIs by reading API documentation and automatically starting to build an integration.
If you had told those two bright-eyed freshmen students if they knew what sort of rocket ship journey was in store for them, there was no way they would have believed you at the time. If Ding could impart one final piece of advice for founders, it would be to pick your partner carefully: “My co-founder will be the man of honor at my wedding. Is that the case for every co-founding team? Probably not. But who you’re going to be working with is a huge decision. And if the company’s doing well, it’s forever. So choose wisely.”