Feature Discovery: Avoiding Wasted Work

It doesn't matter the features you have in your backlog. Every one of them is most likely based on assumptions. That's why, unless you're sure about the value a feature will bring and its format, you should start with feature discovery instead of feature development.

Assumption is a fancy name for risk, and you mitigate it only by collecting evidence. Here are some common assumptions that you need to back up with evidence:

We have a clear understanding of the problem

We mapped the chain of events related to this problem (the context involved)

We know what the user experience of the solution should look like

The customer will be able to understand and use the solution (you'll only have this answer after you build and test a prototype with customers)

We know how customers' success will be verified (metrics)

That's a starting point, but you could also consider other factors, anything that reduces the risk of developing anything detached from reality.

Remember: you'll only be able to collect evidence with experiments and the shape of it can vary a lot. For example, if you need a better understanding of the problem, interview customers. If you want to be sure about the ideal format for the feature, put together a prototype, and test it.

In an ideal scenario, you'll put into production only features backed with evidence. Your end goal is to execute with minimal to no uncertainty.

Prioritizing

Once you're done with feature discovery, you need to decide when to execute it. Again, your goal is to transform assumptions into concrete facts:

We know how much time, effort and resources this feature is going to require

This should be easy, as you'll have a tested and validated prototype by now. So you and your team can understand the feature scope and all the specifications of it.

Finally, to decide how this is going to be prioritized, you should consider the impact the feature will have on both the business and the customers.

Our final assumptions:

The solution is valid once we see customer behavior (___) changing to (___)

This behavior change will create (___) business value

Define metrics and a time frame for the two parameters above and present it to the team and/or the CEO with everything done so far. It's quite a lot of work, but the big picture here will show:

a) Clear reasons why this should be put into production

b) A validated format, so everyone involved knows it will not be a waste of time

c) A precise effort and resource estimation. After all, we outlined the solution.

d) The value it brings and how to verify it

Stop wasting time in uncertain execution. Start investing time in feature discovery.