The Bytes

How We Iterated on 10 Ideas in a Month

After a discussion (that lasted 30 seconds, as it was an obvious decision for us), we decided to share our complete journey with our new startup, Airbyte, with as much transparency as possible. Learnings should be shared! Anybody can leverage our journey, and it’s also a good way for us to take a step back and understand what we did right and what we could improve upon. So here is our first story: how we iterated and sorted through 10 ideas in our first month of working on a new startup.

Michel and I have been very close friends for 7 years, and have always wanted to work together. We’ve worked on 3 side projects together in the past years. But it was never the right timing. I was building my Nth startup or he was in hyper-growth mode at LiveRamp. 

Last September, I was closing the chapter of my last startup, and Michel had just left his company to found his first one. For once, it was good timing for both of us. We just didn’t have any ideas yet. So we started iterating on ideas. But we needed a process to identify the good ones from the bad ones.

When you have an idea, the natural thing that happens is that you fall in love with this idea, and you start to build a very strong confirmation bias towards it. You’re exhilarated; you believe in it so much that you become blind to strong signals against it. But you should definitely NOT go into it blindly just because it’s your idea and you love it! 

We came up with a 2-day process to help us determine whether an idea is good or bad, with as much objectivity as possible. But before diving into the process, there was something we needed to keep in mind. 

It’s about the problem you’re solving, not the solution

You have surely heard the phrase, “Ideas don’t have any value, execution has.” The point behind it is that ideas will be iterated upon and will evolve through the execution phase. Ideas actually often revolve around solutions naturally, but you need to pinpoint the exact underlying problem. Your product will not have any value if you don’t solve a problem for your customers. The question is: Is this a problem worth solving? Your idea might not be worth anything yet, but the problem you’re addressing represents a business opportunity. 

Focusing on the problem will also help you be less biased toward the opportunity, and rethink your solution entirely within its larger context. This also helps you see what alternatives are  currently being used. 

So what happened in our case? How did we not succumb to diving into the idea right away, and how did we take a step back instead? We tried to invalidate our ideas, not validate them. 

Invalidating our ideas intentionally in 2 days

The first thing to agree on if you are several co-founders is that your intention is really about invalidating your idea, not about validating it. This mindset is very important. Otherwise, you will give in to a confirmation bias that you need to fight against. 

Why would you want to invalidate your idea? Well, it’s better not to spend more than 2 days on a bad idea, so you can free your mind for better ones!

Why only 2 days? The goal here is to have a very efficient process that will ensure to invalidate 90% of your bad ideas within 2 days. And that it is very easy to implement that it becomes just natural for you. So it’s not counting hours on any task. 

Day 1: Listing known unknowns

The first day we would start by listing all the assumptions and questions related to the idea and the problem it was addressing. Questions might be:

  • What actual problem are we solving?
  • How do customers solve it right now? (ecosystem analysis)
  • What is the impact of that problem?
  • Who is impacted by this problem? (audience / persona)
  • Can I easily reach those people and, if so, how?
  • Is there any evidence they’re willing to pay for that? How much from current alternatives?
  • Which assumptions does our solution rely on to be successful? What information do we not know yet, and what do we need to know? (known unknowns)
  • Who could give us more educated information on this problem and project? What about our known unknowns?

The first day was spent on research (on Google) to find these answers, and if we felt more and more confident, we would try to find people to talk to for the next day, in order to address our known unknowns. We would ask for a 10-min call – no longer – so it’s hard to say no, because people can always spare at least 10 min. We would send a text if possible, as a text is more interactive and harder to ignore. 

The text might say something like “I have this new idea and I can’t convince myself not to pursue it. Can you give me 10 minutes of your time tomorrow to convince me not to? :)” 

People are a lot more curious and willing to spend time with you if they see that you will listen to them. Asking people to convince you not to do something is a great way to show that you will consider every bit of their feedback. 

Day 2: Verifying known unknowns and discovering unknown unknowns

The second day is mostly about interacting with knowledgeable people. We would try to leave some time between each interaction so we could take some steps back on the new learnings and decide which new questions to ask. 

The goal here is to address your known unknowns that you listed the previous day, and also to go after your unknown unknowns. These people will know a lot more about the industry, and actually anything that can impact your project. So don’t hesitate to ask questions like, “Is there anything I should know that could impact this project?”

Being in San Francisco helps; everybody knows a person working in this industry. We tried to talk to at least 5 people who might have some educated feedback on the idea. Five is not enough, for sure, but at this stage, most of the time 5 will be enough to see any big holes in your idea. 

For instance, in usability testing, we often refer to 5 interviews, because statistically, there is an 85% chance that you see the patterns that exist. It’s the same here. 

Note that if you invalidate your idea at your first call, just cancel the other calls and move to the next idea. I repeat, the goal is to invalidate the idea, and in that case, it’s mission accomplished! Good job! Now switch to the next idea.

If at the end of the day, you are still convinced about your project, you should definitely continue with the invalidation process. It will help you define your solution, the go-to-market, your persona – everything you need to get started –  with a more objective perspective. 

On our side, 40% of our ideas were filtered out after the first day. Fifty percent more were winnowed out after the second day. 

A list of ideas we iterated on

So in September 2019, that’s exactly what we did. And we iterated upon quite a few ideas. Here are a few of them:

  • a lending and financial services solution, catered to immigrants. We would start by offering student loans, in addition to bank accounts and credit cards. The vision is to get rid of the credit score system and citizenship limitations. Our approach would be to partner with international banks to get vouchers from immigrant students’ home countries, to mitigate risks. 
  • a teleoperation product for autonomous vehicles. It will take years to build a fully autonomous vehicle, and experts in the industry actually doubt that it will ever be possible to achieve 100% autonomy in every situation. Teleoperation bridges this gap and accelerates the safe deployment of autonomous vehicles and services, at scale. 
  • an integrated banking, accounting and tax service for the workers of the gig economy, such as drivers (Uber, Lyft, Amazon, Postmates, etc.), Airbnb hosts, Upworkers, Fiverrs and other gig workers (people using gig work platforms, a subset of freelancers). It would have been free, and we would win money through the interchange fees on our credit card. 

We’re not keeping the other ideas for ourselves, it’s just that some were...really not good!! Not even worth mentioning 🤦‍♂️.

Fun story: We actually submitted our first YC application with the third idea in September 2019. 3 days after we submitted the application we discovered that both Uber and Lyft were working on such a solution. We rescinded our application then. And a month later, Uber announced their solution.

Some people compare startups to love stories. Don’t do that. Well...unless you start relationships by trying to invalidate them as much as possible! But I’m not sure it’s the smart choice in that case.

Open-source data integration

Get all your ELT data pipelines running in minutes with Airbyte.