When working with clients you always have to work with estimations. If you do them well, your team will be happy. And your company will run well.

"experience will tell you what info you are missing, but you don't know that you are missing"

Usually, the trick to giving good estimations is a combination of having all the necessary information and experience. The experience will tell you what info you are missing, but you don't know that you are missing. Experience will also offer you a history of similar work and how long it took.

When it comes to missing information, many times we make assumptions. Some of them are made because we are constrained by time. Other times we make them unconscious.

There are times, though, when we know exactly what information we are looking for. And we have two choices: ask the client or find out yourself. Asking the client is, most of the time, the preferred choice.

However, there are occasions when either the client has a really low response-time or simply lack knowledge.

Last Friday, we were up against needing to know if an Android app (which the client has) does or doesn't do something specific. We could have asked and waited 2-3-4 days for an answer. Or we could have tried and figured out ourselves.

At first, I came up with a simple test. Choose a specific section, make some specific changes and observe what happens in the Android app (or IF something happens). I thought it was good enough. My colleague came to me and said something along the lines "yhea sure, we can do that. But how confident are we that this test gives us a definite answer". Basically the question was how confident would we be after we have the results?

And that was the trigger. The challenge. The "oh, let's see moment".

Turns it out she was right. We couldn't have been certain that the results of my tests could have been replicated in other sections or circumstances just as well.

A couple of hours later, after playing around with Android Studio, sniffing network requests, finding the .apk of the app in question I had the answer. Level of confidence? 100%. Happy moments.

Even more interesting, for me at least, was that the final answer was actually in opposition with the assumptions I made. I wasn't interested in being proven right. It is much better to be able to give a final answer in full confidence, backed with data or tests.