Merrill Lutsky Saw Opportunity in the Final Days of Phabrictor; Here’s How Graphite Grew Out of That

Like so many great inventors, Merrill Lutsky and his team created the tool they themselves needed. Working at Square, Oscar, and SelfMade, where I hired him, he saw the challenges of internal code review systems, and saw those challenges heightened as the open-source spin-out of Facebook’s solution, Phabricator, ended support last summer. Merrill and his team created a new tool, Graphite, that now counts Ramp, Datadog, Bolt, and many other large businesses as clients.

I spoke with Merrill on this business venture, knowing when to count your losses and move on, and how to stay close to the product.

Tell us about what you're building at Graphite.

We're building a code-review platform for fast-moving teams. For anyone who's trying to ship software quickly, whether it's a small team that's just starting to formalize the process of code review up to a larger company that has many teams and complicated approval rules, we want to be the place where code review happens.

How did Graphite start?

Graphite started as an internal tool we built to solve our own needs around code review. My Cofounder Tomas and the first few engineering hires we made all came from Facebook and had experience using Phabricator, the tool and the set of tools that Facebook has built internally around code review and around creating changes.

We were working on a different developer tooling idea at the time and what we found was, as we were hiring a few engineers, we quickly got to the point where the folks that we were bringing on were noticing their productivity declining in the absence of the tools and workflows they were familiar with. It was so much of a pain point to them that they actually took it upon themselves to build the first version of Graphite.

So you had your engineers build something internally and now that's become the company. What was the process of realizing that that was the right move for you?

Last summer, the sole maintainer of Phabricator, the open-source project, announced that he was ending support for it. So it created this interesting conversation on a lot of forums of former employees at Facebook, Uber, Dropbox, Pinterest, Quora, and any other company that was using Phabricator internally. A lot of folks in their alumni communities would start asking things like, “Hey, is anybody building a tool like Phabricator that enables the stack changes workflow that encourages my engineers to write smaller PRs and not be blocked on code review? Is anybody doing something that solves this need? Is anybody going to maintain Phabricator? Is everyone just moving to GitHub? Where's the community going after Phabricator shuts down?”

We saw that and we realized that we’d actually built at least parts of what these engineers were asking for internally. We decided to reply to some of these discussions and see what happenedI think we shared a quick survey and short videos of what we’d built just to gauge interest, and the response was overwhelmingly positive.

Talk to me about the shifting role from being CEO of a developer tools business that's trying to find its real wedge to now finding it. There is now a completely different set of problems that you're working on. What are those new problems? How is the transition going for you?

Before we had confidence in Graphite as our company's direction, much of my focus was on the product:figuring out what features to build, working with the team to make sure that we're iterating quickly and communicating our progress to users, and then collecting and integrating their feedback. Now, most of my time goes towards building the team and thinking about the company as the product. I essentially work at a different level of abstraction now—instead of working on the product directly, I have to build the team, systems, and culture that creates the end product.

Initially, it can be a little scary to think about that big of a shift in your day-to-day responsibilities. For me, coming from a product background, this framework of thinking about the company as the product I’m building rather than managing the product directly  gave me a good perspective on how to approach that challenge.

Would you say that you're growing through product-led growth? How would you define product-led growth?

We're in a fortunate position in that a lot of our growth has been organic —we've only really done publicity around our waitlist launch and then around Series A. We've been on a few podcasts since then, but most of our new sign-ups are engineers at companies where they already have a colleague using Graphite.

This highlights a really nice property of code review—it’s a fundamentally social activity, which lends itself super well to a bottom-up, product-led growth motion. Engineers review each others’ pull requests every day, so you have these natural moments where they can share Graphite with their peers. Because we sit on top of GitHub, all it takes is one engineer at a company to adopt and get value out of Graphite, and then they can champion us to their team organically instead of having to convince everyone up-front.

We’re reinforcing this by building features that work best when the author and reviewer are both on Graphite, so users are naturally incentivized to spread Graphite to their team to unlock more value from the product. We also invest a lot of time and energy into our Slack community and docs to help teams scale their usage of Graphite. Once we see that a company is highly engaged on Graphite, we can reach out to provide them with more direct support, and then eventually start the conversation around our enterprise offering.

What's the metric that you're trying to move most right now?

The biggest thing we're looking at, and one of the key challenges of code review, is the activation funnel. We’re really trying to optimize when a new user comes to the Graphite site, signs up, goes through onboarding, and starts submitting, reviewing, and merging PRs with Graphite.

The biggest challenge in the funnel is that we have to make a pretty big ask up-front—to get access to your company’s source code— before we can really demonstrate value. There’s a ton of trust-building for us to do here; we have to educate new users on why we need this access and what we’ll do with it. We also have to emphasize how heavily we invest in security—we’re in the process of obtaining our SOC2 certification to help speak to this.

There's also a big educational piece there. We’re promoting a new development workflow and we have to show new users the benefits of stacked changes, and then guide them through making their first stack, reviewing their first stack, and merging their first stack. Users who make it through that loop once are far more likely to retain and spread Graphite to their peers. The biggest thing for us is just to optimize that activation rate so that we can then unlock more growth efforts.

What are you doing to stay close to the product?

We recruited the first 50 alpha users around this time last fall by responding to discussions that we saw on these ex-Facebook alumni groups. Whenever somebody would post about Phabricator, we'd follow up with them and anyone else on the thread who was interested. We sent them a survey about what code review tools they’d used in past roles and what features they thought were most important, and then once we were ready we invited them to test the very first version of Graphite.

One of the things that helped us the most in the early days was putting the first 50 users into a shared Slack channel. We would post in the channel we released new features, and our alpha users would share feedback and bug reports. I’m really grateful for that initial user group and for our users today who continue to give us so much valuable feedback on Graphite.

We also had a standing weekly Zoom call with almost all of our alpha users. It was really helpful to get to know them and understand what they were working on and what their challenges were, both in the context of Graphite and more generally.

How did you get people onto that call?

When we gave those first 50 users access to Graphite, we made taking regular feedback calls one of the conditions. We said, “This is something that's in development and we're not charging for it. Our ask instead is that you pay us with your time and tell us what you think.”

I think this was good both because it gave us a lot of great feedback, but also because it was a commitment mechanism that made everyone in the alpha program take it seriously. The fact that users were willing to commit a large amount of their time in order to use Graphite was strong validation that we were building something they needed.

How good was attendance?

Pretty solid. I think that was one of the things that was surprising and really encouraging to us. Even today we still have regular feedback calls. We can’t do them weekly anymore, but we have regular check-ins with champions at many of the large companies that use Graphite. We now have a dedicated Slack group for the Graphite community with over 6,000 engineers, and they give us really valuable feedback on a daily basis.