If you have a vision and a strategy and if everyone in your team can articulate it, you are actually pretty amazing. Most companies don't have that, they struggle with the strategy. The roadmap articulates how you are going to achieve that strategy from where you are today. I think of it as layers: first the vision, then the strategy and then the roadmap.

I like to frame the roadmap as a prototype of your strategy. It speaks to how you believe you're going to achieve the strategy, but you're going to learn in that journey. It's a prototype because a prototype is something you can change easily, whereas a vision is a long-term thing that you don't change often.

If you want to build an investable startup, it should be a significant size and you should be pretty passionate about the problem you’re solving and the segment you’re serving. When you don't have enough passion or you can't identify with your market segment, there is a tendency to drop off in that journey.

I don't think that in the early stage profitability is very relevant, to be honest. It’s more the market size and retention and the passion that you have for it.

The idea is that if you're doing discovery properly, you're killing a lot of the ideas there so that only a few ideas come out of discovery and go into the delivery track. As a result, you should see less failure and you should see happier developers.

For me, customer research and product analytics are all important feedback mechanisms from your users and customers. It's also a key feedback mechanism to influence your strategy. So you should be absolutely doing it every day.

Minimum viable product is a term that came out of The Lean Startup by Eric Ries. From his perspective, it's the smallest experiment that you run to test a hypothesis. Minimum viable product can also for some people mean that it’s the first version of the product that you're actually going to build. It's not an experiment. 

The method of collecting feedback really depends on what you’re testing. It’s about what assumption you want to validate and using the right tool or experiment for that.

There’s no magical number. I find that teams also end up in a little bit of analysis paralysis and get lost in that phase. The B2C space is very difficult. For some of these tests statistical significance is important, and that's why sometimes partnerships also make sense, especially in the B2C space where, for example, corporates have huge customer bases.

Firstly, you have to create a product that's valuable for the users who are using it because it's solving a problem. Secondly, the product is usable. It's not just solving a problem, but people can and do use it. Thirdly, it's feasible, so you can build it and from a tech perspective there's no feasibility challenges. Finally, it's viable.