How to build a successful DevTool Company and what to avoid
No matter how much money someone invests, here are the early signs to watch out for
In the start-up land, rules are meant to be broken. Every time we create rules for how something is supposed to work, someone else breaks the stigma and turns the world upside down. And yet, there are very strong patterns. One of them is: what can and cannot be sold to developers - no matter the hype, how many millions a start-up raises, or the number of billions in its valuation.
Early Looker was a DevTool too
This past week
from FirstRound posted an interview with Lloyd, Founder of Looker (someone who took a chance on me back in 2014, for which I will forever be grateful). It is a phenomenal interview and definitely one worth watching ⬇️. As usual, Lloyd Tabb does not disappoint. You would think the man would retire by now and disappear into irrelevance. But he has other plans.In the interview, Lloyd talks about his almost-instantaneous Product Market Fit by leveraging consulting as a way to solve customer pain points, and remove friction around trying to sell the tool.
But there are a couple of subtle points that are missing here from the context…
The early version of Looker was essentially a DevTool. The very thing that people likely associate Looker with today (i.e. Visualizations) were not in the actual early version of the software. Surprising, no?
Before you go on telling your executive team to drop all visualizations, and tell your sales team to start selling consulting services instead of software, you should really pay attention to the next few points.
For almost a year, this is what Looker’s homepage looked like:
Things got a little bit better the next year, but not by a lot. The biggest two improvements: Logo and Call to Action.
As for the buzz, this movie cover was basically the state of hype surrounding Looker for the first few years - if you went on the internet and typed “looker”, all you would see is a reference to this cover (super weird, if you ask me). That is to say, unless you were an insider, you probably did not hear about us.
Basically, it was not until like 2-3 years later that the company started to grow legs and its marketing started to take shape. Here is yours truly, in 2015, when we launched something in marketing that would end up having a much more significant effect on the whole data industry than most realized at the time—and maybe no one realizes still—but more on that later, perhaps in another post.
Developers, Develo… —I mean—Expectations, Expectations, Expectations…
If it is not clear yet, I mention all of this because technical audience GTM kinda follows a pattern. And the pattern is basically to keep everyone’s expectations extremely low until you can deliver on them. This is what Looker excelled at!
the pattern is basically to keep everyone’s expectations extremely low until you can deliver on them
Every customer interaction and engagement at Looker followed a pattern:
people are confused,
they see some signal of credibility,
they are curious,
they engage,
they are astonished,
they turn into loyal fans.
It turns out this works well because technical audiences - whether developers, or, as in the Looker’s case, data people - tend to be a skeptical bunch. They are skeptical of anything that promises them the moon. They are skeptical of good copy and clever sales tactics.
Of course, this does not necessarily mean that a technical buyer is a more sophisticated buyer. It just means that when dealing with a technical buyer the delta for exaggerating your product’s capabilities is very small (and it is even smaller for DevTooling in data).
when dealing with a technical buyer the delta for exaggerating your product’s capabilities is very small
AI and exaggerating capabilities
Enter “AI” - the world in which almost everyone exaggerates existing capabilities, so what do we make of it?
The answer is complex. On the one hand, of course entrepreneurs are going to exaggerate their current capabilities and describe impossibly large visions. And maybe that’s OK. Investors are an experienced bunch, and know how to level down expectations - sometimes even too much.
However, what happens when that enthusiasm from an investor pitch carries over to how the company pitches itself to Developers?
It turns out this can be a major problem:
When investors get involved in hyping up a developer-focused product, that is a major problem. That delta between current capabilities and the promise becomes unrealistically high; while OK for Uber and Lyft - the same type of hype creates a total destruction of value for any developer-focused product. The trust is lost forever, and few actions will ever turn things around.
The trust is lost forever, and few actions will ever turn things around
And while it is easy to say, the market is big - we can always find other users, that is somewhat misleading for the following reason:
DEVELOPER MARKET IS ONLY AS BIG AS THE NUMBER OF SENIOR VISIONARIES THAT BUY INTO YOUR VISION.
Let’s imagine you’ve screwed up that initial launch and lost all trust with the senior users, but you are still able to win over all the juniors - great, right?
Not so fast! What happens next is the difference between:
(A) category—Google’s $2.6B acquisition of Looker (on $280M raised) OR Microsoft’s $7.5B acquisition of Github (on $350M raised)
AND
(B) category—Google’s ~$12-20M acquisition of Kaggle ($16M raised).
Companies in B are successful at building large audiences, but those audiences are more of an expense than a real way to generate any revenue. A fresh graduate from a computer science department might be super excited about your product, but what difference does it make if it is going to be another 5-10 years before this new user commands any kind of budget. Only large companies can afford such a strategy, which leads to acqui-hiring, not real M&A.
The AI-creation curve
Every great recommendation system depends on who the early adopters are for its initial product. Netflix today only has a great recommendation system because its early adopters were religious about leaving feedback - good and bad - on the movies they watched even at the time when the experience of watching films on DVDs was disconnected from the Netflix platform itself. By contrast, what’s the percentage of people who took the time to leave reviews for Blockbuster-rented films at around the same time - probably minuscule.
Such examples of early adopters are endless in tech and explain a lot of success and failures as companies attempt to generate scale. Focusing on poorly engaged early adopters eventually leads to bad recommendation systems and ultimately failed products.
However, what happens when a DevTool’s early adopter base are still loyal users, but very junior ones? Basically this depends on the type of DevTool.
If the tool is a teaching tool. That tool’s AI can and should probably benefit greatly from learning all the basics around its users’ questions and gaps in learning. For example, an onboarding tool could be really good here. A new junior developer joins a firm, and the tool walks this new developer through the codebase. That could work really well.
However, for more advanced tooling - like an “AI Engineer”, “AI Data Engineer”, “AI DevOps” or “AI Web Developer” - AI has to be trained on good inputs. Just like in the Netflix example, the tool will need its own version of movie buffs to help curate what’s good and what’s bad about AI’s responses. The GTM is everything in this space.
Final thoughts
Enough of the bad. How should such situations be avoided altogether?
Aside from NOT valuing pre-revenue companies in the billions, there are many other ways that investors should avoid getting involved - even if done under good intentions.
AVOID: A typical scenario might look like this: to get some community goodwill going, an investor might host the start-up’s hackathon. Seemingly innocent action, but one that sets all the wrong expectations. Instead of early adopters who are actively seeking a solution to a particular problem, our hypothetical start-up gets a bunch of (often unemployed) kids who are very excited about a particular VC firm.
DO THIS: The right thing to do instead is to help the said start-up recruit an internal community leader and / or evangelist.
AVOID: Pushing your start-up to launch on platforms where everyone launches (producthunt, ycombinator, etc). While these do drive initial traffic and github stars, ultimately the traffic is short-lived. The attention is there only until the next thing gets posted.
DO THIS: Help your start-up with marketing hires, so that they can build their own distribution channel.
This is obviously a very long list, but you get the idea. Bottom line, investor activism is truly the wrong thing when it comes to Developer Tooling - and maybe that’s a good thing.
Finally, I am clearly quite passionate about this subject. Yes, I worked at Looker and saw what success looks like. But incidentally, I also run an AI DevTool start-up myself, so this subject hits close to home. Having said that, I hope the above ramblings were at least a bit useful.
Have a good weekend everyone!