DevTools Marketing with Jason Lengstorf
Jason Lengstorf thinks Vercel is both a great example of devtools marketing and a cautionary tale of the same thing taken too far — selling an identity so hard that using the product without loving it starts to feel like you're not welcome.
Jason Lengstorf is the founder of CodeTV and previously led developer marketing at Netlify. This is his second time on the show, and the conversation goes deep on why constantly reinventing your messaging every time you learn something new is exactly the wrong instinct — and why consistency, not cleverness, is what actually builds trust with developers.
We get into why Gatsby's community evaporated the moment it stopped being the only good option, what Astro and Warp and Cursor are doing right in their positioning, and why he thinks video is the highest-leverage thing a devtools company can produce — including the actual mechanics of turning one recording into a dozen different marketing assets.
The Q&A below is experimental — extracted from the episode transcript with AI. If you spot a mistake, let us know at jack@scalingdevtools.com.
Why is it a mistake to rewrite your messaging every time you learn something new about the market?
The instinct comes from a good place — you want the strategy that's most adapted to what you're hearing. But what makes it possible for someone to care about your brand is continuity. If a friend showed up every time with a brand-new hobby and a completely different group of people, it'd be exhausting — it doesn't mean you like them less, but you don't know what to expect from them anymore. Companies are the same: if your positioning keeps contradicting itself, I can never get a clear picture of how you fit into my workflow. And remember the asymmetry — you spend twenty-four hours a day thinking about your company, but as a consumer I spend maybe fifteen seconds thinking about it, if you're lucky. If that fifteen seconds feels inconsistent, I write you off as irrelevant and go with the thing I already understand, which is often a big, boring, predictable incumbent like Jira.
How long should a company commit to one message before it's allowed to change it?
At least a quarter, ideally six months. If you're consistent with one message for three months, and you're positioned properly, the most active people in your community will start repeating that message back to you unprompted — that's the sign it's landed. Give it six months and even people who aren't your biggest fans will have internalized it too. Big brands understand this: GEICO's gecko and Budweiser's Clydesdales have been the face of those companies for decades, not because those companies can't afford a rebrand, but because they know a consistent message beats a perfect one. Internally there's chaos, new features, decommissioned products — all of that is happening. But externally the message stays exactly the same, which is what makes them look stable and trustworthy. Smaller devtools companies externalize their internal chaos by changing what they say all the time, and that's the mistake.
What's the risk of marketing on specific technical benchmarks — being 30x faster, for example?
Anything technically bound can be proven wrong by a competitor's benchmark, and it evaporates the moment the underlying tech becomes a commodity. When every JavaScript framework was competing on being 'blazing fast,' they all eventually became Vite plugins, and now everyone has roughly the same performance — so what's your differentiator now? The same thing is happening with AI: if your entire premise is 'we give you AI,' that's not a business, because any company can get that by paying OpenAI or Anthropic. Coca-Cola doesn't market its bottling technology — when they switched from glass to plastic, that's not the message. They're selling nostalgia, a state of being, a feeling of being on a beach on a hot day. If you articulate the actual outcome for the user clearly, the underlying technology can shift a hundred times and your marketing doesn't have to change at all.
What happened to Gatsby, and how does that contrast with what Astro is doing?
Gatsby had an incredibly strong community, and it felt like everyone was building everything with it — but I think the founders were genuinely surprised when people didn't stay loyal once real competitors showed up. The community existed because Gatsby solved a problem in a way nothing else could; once other tools solved it the same way, there was no reason to expect loyalty, because there was never a monopoly on the underlying utility. The bigger mistake was investing in Gatsby Cloud as if all Gatsby sites would always be built on the Gatsby build system — so once it stopped being fashionable, migrating off felt like a genuine pain, and usage fell off a cliff. Astro, so far, has done the opposite: it deploys anywhere, works with any framework or none, and it's standards-based, so my skills transfer even if I stop using it someday. My loyalty to Astro isn't about identity — it's that it's the thing causing me the least friction to get the outcome I want, which is a much more durable thing to build a business on than being someone's whole personality.
Is there a risk in building too strong an identity-driven community, the way Vercel has?
Vercel is a great example of doing devtools marketing right, and also an example of it going a little too far. When a community starts to feel like it's built around ousting people who dissent, that feels bad — you're either all the way in, or you're treated like you're not really a member. Some companies manage to be a lifestyle without it becoming a required personality; Vercel, I'd argue, has crossed into requiring it, so that if you're not all-in, you're implicitly a hater. Most devtools are the equivalent of a toothbrush — I don't assign my identity to my toothbrush, or the brand of socks I wear. If your whole community strategy depends on people making your tool part of their personality, remember that's fashion, and fashion changes every eighteen to thirty-six months — that's exactly what happened to Gatsby.
You've said developers aren't leads. What does that mean for how devtools companies should talk about their product?
What's most relevant to a customer is never your specific technical specs — I have never in my life cared how many gigaflops of processing power something has. What I care about is whether picking up your tool gets a task done in two hours instead of four, and you need to be precise about what that thing is, and honest that it's not for everything. Warp is a good example: they didn't sell 'a better terminal' in the abstract, they showed you can click and drag to highlight command output and copy it instead of hitting backspace repeatedly, or right-click to copy an error instead of dragging to select it. That's small, but materially valuable to anyone who lives in a terminal. Cursor did the same thing with AI coding — the pitch wasn't 'we have AI,' it was 'you get a diff instead of copy-pasting AI output back and forth,' which is a specific, clear outcome: integrating AI into your workflow without turning you into a miserable copy-paster.
Why do you think video is the highest-leverage format for a devtools company?
One recording turns into an enormous number of assets: cut it into shorts, drop the audio as a podcast, turn the transcript into a blog post, turn that into a tweet thread. When I was at Netlify, we'd been doing conferences, event booths, and paid keyword ad spend, and the ROI just wasn't what it used to be — and post-2020, meetups largely disappeared as an option too. Meanwhile everyone was sick of watching another Zoom call, but people still watch a lot of TV, so the question became what lessons we could borrow from why people choose to watch something: real stakes, real stories, tasks relevant to their interests. With something like my show Web Dev Challenge, you get the show itself as an asset, the apps built during it as another asset, a relationship with the developers on the show, social clips you can run for weeks afterward — every piece points back at the others and keeps generating new things to share, instead of you having to remember to go talk about something you did once.
How do you think about the attribution problem — proving that this kind of content actually drives revenue?
You will almost never get clean attribution, and that's fine. Someone watches a video, tries the tool a little at work, shows their manager, the manager decides to look into it for the company, and eventually there's a sale — that's five hops, and there's zero chance of clean attribution on any of them. But if you have any real connection to your community, you can feel it working: the way people talk about you, the way they reach out, feeling like they're part of something with you rather than being sold to. Netlify put real money into keeping the Jamstack community's local meetups running — functionally like running a nonprofit around that community — and you can draw a pretty clear line between when they stopped making that investment and when they stopped showing up in the conversation. You can't put 'community trust' on a spreadsheet, but if your ACV is $250,000, you don't need many of those five-hop sales to pay for the entire investment.