In my last post, I suggested turning your riskiest assumptions into hypotheses that you can test. But how do you tease out the assumptions you are making? It’s a tricky question. We make a great number of assumptions every moment we are alive, on topics ranging from physical laws to the likely behavior of others around us. The biggest challenge is that many of them are assumptions we don’t even know we are making. We need a tool for finding them.
I recommend “looking behind the why”. Here’s how it works. If you are about to take some action (new feature, market segment change, etc.), you are doing it for some reason (the why). Frequently, that reason is unspoken, so the first step is to state it explicitly. Once you’ve done that, you get to the second step: how do you know that what you are doing will achieve that? There’s a pretty good chance that you are relying on assumptions rather than empirical evidence. The formula looks like this:
- Why am I taking action X?
- How do I know that action X will achieve that?
For example:
- I am modifying a feature so that a given operation takes less time.
- Step 1: Why am I modifying it? Because lowering time taken will increase usage.
- Step 2: How do I know that lowering time will increase usage? Um…I don’t. I assume that the reason usage is low is that it takes too much time to use the feature. Ah! There’s an assumption I should test.
In reality, the Q&A won’t be so clean. You might have to dig deeper behind those whys to find the hidden assumptions, but they are there! For example, the same conversation might go like this:
- Q: What are you doing?
- A: I am modifying a feature so that a given operation takes less time.
- Q: Why are you modifying it?
- A: Customers said they weren’t using it because it took too much time.
- Q: Hmmm…sounds like a trap. Remember: what the customer says and how the customer actually behaves are two completely different things.
- A: Oh, well I can understand why the time taken could be a problem in their context.
- Q: And how do you know that fixing this problem (making it faster) will actually change their behavior (using the product)?
- A: Um…I don’t. I assume that they would use it more if the operation took less time.
- Q: Ah! There’s an assumption you should test.
There could also be multiple reasons you are doing something, so be sure to repeat this to tease out other assumptions: Step 1: Why else?; Step 2: How do you know?
An especially good place to find these is in the “so that” clause of the common “As a…I will…so that…” story pattern. This clause should express the value the story provides to the user. It is a good first stab at answering the first question: why you are doing the story? Step two is to ask: how do you know that completing your story will achieve that end?
This may seem tremendously simplistic, but I’ve found it to be incredibly powerful in teasing out assumptions I am making unconsciously. Remember: look behind the why. Put differently: How do you know that what you’re doing will achieve what you are hoping?
Next up: how to build truly lean experiments that test your hypotheses without just building the whole product out.