Heuristics for the Data Scientist Entrepreneurs
This is a lot of inside baseball.
The motivation is to share information while acknowledging that it’s wildly anecdotal.
It’s directed at data scientists thinking about business.
Andrew and I founded Authintic in late 2012. We landed three great customers. We met between 1,600 and 1,900 well wishers, competitors and prospective customers. Five major market hypotheses were tested. Revenue was earned and value was generated. Authintic was acquired by 500px in early 2014.
Thrilled. Very excited.
And a tad skeptical about the lessons learned.
People are terrible about extracting causal factors from an experience. I’m people. So I reckon that applies to me too.
A sample size of 1 isn’t authoritative. It doesn’t constitute proof, or evidence of a natural law. I can’t and won’t pander about Ferrissisms.
There is value in experience.
I understand how my actions map back against strategy and the justification for most of those actions. I know this because of the magic of the structured checklist and careful goal alignment strategy.
On Evidence-Backed Checklists
I believe that executing an evidence based checklist worked. I was punished whenever I deviated from the checklist.
Things went better when I stuck to an evidence driven checklist.
- “Find a business cofounder” is on that checklist.
The Genome Startup Report says “that balanced teams with one technical founder and one business founder have 2.9x user growth and are 19% less likely to scale prematurely than technical or business-heavy founding teams.”
I can validate the value of a balanced team. It worked. Having a business cofounder is immensely valuable. We always joked about an ad exec and a data scientist walking into a bar. That worked.
- “Seeking and Accepting advice”
Seeking and accepting advice also seemed to work. I met a lot of great people through Digital Media Zone, and NYU Poly and at General Assembly. We also got terrific support from the CDMN. I can genuinely say that some people added value. And I thank them all.
Founders that seek and accept advice have greater odds of success.
That worked. That’s confirmed.
- “Count noses”
Before one single line of code goes down, before one single letter in a piece of sales material goes out, you better be able to identify a handful of people that would want what you’re solving. Counting noses is absolutely integral in producing empathy in design. Early feedback is best feedback. That seems to have worked.
I was very efficient in rejecting market hypotheses.
I may have been too optimized.
It’s impossible to tell because all seemingly bad ideas all starved to death in the cold.
On Strategic Alliances
When does 2 + 2 = -9?
When startups try to work together, of course!
I had terrific collaborations with several technical co-founders.
I had several experiences that really were not good.
There’s a lot of circlejerkery involved whenever founders meet – and I get the impression that offers to work together are like hand shakes. Obligatory. I just can’t figure out if there’s a culture of empty promises where we have no real intention to follow up.
Daily circlejerkery is an activity that may deceptively appear to be productive, but it is not. It rarely moves you any closer to truth. It doesn’t get you anywhere close to market validation. (You could run a business dedicated to simply stroking egos, and do very, very well).
When I did consider partnerships, I actively looked for the immediate product implications. Did it fill a need that I had in my stack? I always explored technology from the perspective that I would pay for access to that technology and use it. If I couldn’t justify the market, I wouldn’t build from it.
Certain ecosystems refused to give any indication on their intentions. That, in turn, caused mistrust and terminated the discovery process early. These were the easiest conversations to end because there was strong evidence that a single line of code would generate zero returns.
And it is entirely likely that I ended several of those conversations very early.
The demand for market validation, an actual nose-count, before a single line of code went down, likely aborted quite a few avenues. If I couldn’t count a single nose, then any prospective market just might as well not exist. The actual question of whether it does exist or not is irrelevant. I really don’t have the time to find a new market.
And the set of ideas that people have always exceeds the set of ideas that you can actually test.
In retrospect, I would be far more stringent in accepting meetings and far less stringent about counting noses.
An extra two hours of cold calling may have yielded more signals.
It requires a checklist of at least three criterion.
Think critically about those criterion.
You need to plan for wearing multiple hats. Accountant. Lawyer. Coder. Architect. Statistician. Receptionist. QA. Speaker. Product manager. Project manager. Designer. Executive. Copywriter. Account director. Douche. Coordinator. Form filler.
Among many others.
Many of those hats will make you look very bad. Embrace the suck. Plan for the suck.
Time is your power.
What worked for me was sitting down with my calendar on Sunday morning, and blocking out time throughout the week for different task modes. I cancelled events that didn’t make sense. I sent out invites for those that did.
What made it harder were icebergs in the schedule.
It was not uncommon for a given meeting to move six times over the course of eight weeks. You need to be flexible with that, because it is reality.
People are continuously running late.
Invest in a light laptop and always carry it around with you. You can recover upwards of 15% of your day, using sucky time to wear sucky hats, just by working around the people who are running late.
On New York versus Toronto
In New York, a prospective customer would tell me ‘no’ on the first meeting, and provide me 3 references to people who could use our technology.
In Toronto, a prospective customer would tell me to come back nine times, providing no usable market signal.
A firm ‘0’ is useful data.
Something between ‘0’ and ‘1’ is worse than useless.
Sometimes getting to no is just as hard as getting to yes.
It’s a serious defect in the Canadian market and it needs to be hacked. If you figure out how to do this, let me know.
Your ethics will be tested. You will be tempted. Hold fast.
It’s impossible to say whether your experience will be remotely similar.
Data scientists, and really any analytics professional, have a pretty good feeling for just how stacked the odds of an exit of any form really are.
It’s likely why I met so few data scientist entrepreneurs. And those that I did meet were a very fun bunch – completely aware of our insanity.
Checklists, seeking advice, healthy time management, and progressive hypothesis testing were all high yield activities. 10/10. Would do again.
Strategic alliances and circlejerkery were incredibly low yield activities. And low yield activities are dangerous to an early-concern. 0/10. Would minimize.
I am uncertain if relentless nose-counting is a strategy that hurt or helped. I don’t know how much toxic revenue I avoided. I don’t know how much good profit I left. If you figure out how to calculate this, let me know.
Being Authintic has likely changed my brain in ways that I’m not even aware of.
Being your own company is likely to change your brain in ways that you will not even be aware of.
It may be right for you.