How to become a learning machine

by Rajat Paharia ([email protected]) 94 views 

So you have an idea. A genius idea. A wonderful, brilliant, genius idea. You’re going to build a product or service that’s going to change the world. You toil for months or possibly even years. Every waking (and the occasional sleeping) hour is spent trying to force this dream of yours into existence.

The process is challenging but also intoxicating. You do your market research, make your business plan, build your product or service, and then finally, it’s ready (enough).

Congratulations. On the big day, you launch. The angels sing.

And then nothing happens. The world collectively yawns and continues on its way.

I remember this day from my first company very distinctly. As someone with a natural affinity for building things, I had believed that the formula for a successful product was 90% “build a great product” and 10% “sell it.” On that day, I realized that I had it exactly backward and that the world was a graveyard of great products that nobody knew about. It was a sobering and depressing realization.

The first thing to understand is that you’re not alone. Nine months after launch, Pinterest had a whopping 10,000 users. Beanie Babies were launched in 1993, and sales were poor for two years. Between 2018 and 2022, OpenAI released GPT-1, GPT-2, GPT-3 and GPT-3.5, and most of the world had never heard of any of them.

The second thing is to start learning and iterating as fast as you can. Ideally you launched sooner with an MVP (minimum viable product) instead of later with an FBP (fully baked product), but in either case the good news is that you now have a “prototype” that gives you and your customers something to react to and iterate on. You have successfully left the sunshine and rainbows of thought land behind and are now driving through harsh reality.

Rajat Paharia

This prototype is the whole package that is your business: your product, your marketing, your ideal customer, your sales and your customer experience. And with this first prototype in hand, your job now as an entrepreneur is to become a learning machine: to create hypotheses, test them, learn from the results, make necessary pivots, update your prototype, and then repeat the process until either you’re successful or you decide to call it quits.

Since your resources (time, money, partner’s patience) are typically limited, you need to be smart about figuring out which hypotheses are the most important to test, and you need to figure out how to run those tests as quickly and cost-effectively as possible. In this phase it’s tempting to focus on quick, easy wins, but the goal should be to tackle the biggest risks and problems first. Because if you can’t address those, nothing downstream really matters.

I often think of this process like adding floors to a building. Every prototype you build and test moves you up a floor, and now you can see more of the surrounding landscape. You have a better sense of your product, the market, your customers, and where you fit in, so your next prototype will be even better and move you up another floor.

Learning machines also talk to as many people as they can. First, because selling your product vision to others helps you to refine your own thoughts about your product, it forces clarity. Second, because it increases your surface area for luck, you never know where a great idea, a breakthrough insight, an important connection or a game-changing opportunity will come from.

So, if your product isn’t immediately successful, don’t give up hope. The fact that you’ve heard of all the companies above is because they eventually found their way to success by becoming learning machines, and you can as well.

Editor’s note: Rajat Paharia is the founder and learning machine at Ask Steve, a Google Chrome internet browser extension that allows users to add AI capabilities to websites. The opinions expressed are those of the author.