Reflections on a startup

For the last 18 months I've been building a startup, but we've decided to close it down. It was a tough decision, but the right one. Even though it failed, I have only minor regrets. The company failed to grow, but I experienced personal growth that would be unachievable in a regular job. Would I do it again? Yes.

Cherry Recommends was a SaaS platform (not really, but the term's in vogue) for improving the effectiveness of promotions in e-commerce. We used machine learning algorithms to target individual customers with "the best" promotion. When used in properly, Cherry could increase revenue (yes, revenue) by 15%. Honestly, we thought it was pretty cool. It was ugly, it was hard to use, and hard to explain - but with financial outcomes like that, we were confident we'd find a way. But after 9 months of searching, we just weren't able to acquire new customers at anything near the required rate. We tried changing the our target customer, changing our brand, changing our messaging, and changing our pricing. But those things weren't the problem. The problem was simple: people did not want to buy our product.

What was wrong? If using the tool produced good financial outcomes, why weren't businesses lining up for some of that "secret sauce"? I've been reflecting on this question a lot, and I think there are several related reasons.

There was no 'magic moment' for users. We tried for a long time to find and build one, but it was always a too far into the journey or too hard to find. It's hard to make a magic moment as an afterthought - they must be integral to the product experience. Building a magic moment was hard for a complex product, and I think this was a fundamental mistake on our part. We built a complex product because we were addressing complex problems across multiple businesses, each requiring suble differences in configuration and capability. Many new customers, when onboarded, brought new features with added complexity. Designing a product requires the balancing of competing demands. Customers will ask for new capabilities, but if (as we were) you're pre-product-market-fit those capabilities aggregate like barnacles on a ship's hull. As they accumulate you sail slower and become ugly. In hindsight, adding features and increasing complexity is a bad strategy if you're pre-PMF.

Was there a root cause of these mistakes? We were smart founders with experienced people helping us, but we couldn't make it work. If I had to pick "one mistake to rule them all", it would be that we chose a market we didn't deeply understand. We were software engineers, management consultants, mathematicians, and engineers. We didn't know much about marketing or e-commerce. We did not have founder-market-fit.

So here I am, post-startup, reflecting on the past 18 months. Though the company has been annihilated, I've learned a few things that may prove valuable. Closing a startup is a tough decision. Being a founder means having irrational optimism that you'll succeed against the odds where others have failed. If you closed the business after 1 setback, you'd never make it; but how about 10 setbacks, or 20? At some point, your sceptical mind must intrude and ask the difficult questions. Does this have a chance of working? Can I think of a plan to grow this business? In the end, I believe you must face the music and make the tough choices.

Something that helped us understand our situation was comparing ourselves to other companies in our YC batch. In a batch of over 400, some businesses were growing fast and kicking goals, while others were languishing in pools of uncertainty. We were definitely in the latter. I know it's not always fair to compare yourself to others, but you can bet everyone else is. Investors, potential employees, YC group partners, batch mates - all of them look at your company and compare it to the others they know. Here again, the internal battle between irrational optimism and cold skepticism is waged.

Though I failed to achieve the kind of startup success that makes the newspaper, I achieved personal growth that would be impossible in most other situations. Next time try I startup, my top tips to myself would be:

• Start fast, fail fast. If you can't see consistent growth in 3 months, then stop.
• Simplicity is paramount. Complexity is the enemy
• Wait for PMF and growth before going all-in

For 18 months people gave me advice on how to run a startup. Some was useful, some was not, and some I thought was useful was actively detrimental. For anyone reading this, remember: learn from my journey what you can, but your journey is your own. Nobody else can make the hard decisions for you. It's up to you.