The first pass of constraint identification is aimed at alignment. It only describes where the hot spots (biggest bottlenecks) reside in your customer factory. The next step is conducting some analysis to better understand root causes and get to why.

Here is a 3 step process for conducting this analysis:

1. Run Five Whys

Five Whys is a root cause analysis technique where you start with an undesired outcome, like prospects did not buy our product, and retrace their steps backwards. At each step, you progressively ask why and postulate/search for possible explanations.

Example of 5 whys

Example of 5 whys

It’s perfectly okay to start by postulating a theory for the undesired outcome but remember you have to make an evidence-based argument for your proposal. Whenever possible, back up your theories with quantitative metrics and/or qualitative insights.

2. Metrics analysis

The Customer Factory helps you benchmark the macro metrics in your business model. In order to get to root causes, you often need to dive deeper. Each macro step (AARRR) is made up of a series of micro sub-steps. Analyzing these micro sub-steps helps you further pinpoint where users are dropping off which can shed some clues as to why.

That said, like our constraint analysis from earlier, metrics analysis only tells you what is happening, not why.

You can:

3. Get an early start on discovery

Discovery is an insight gathering activity. It is typically done through customer interviews where you lead the interviewee through a series of open-ended questions and/or observe how they perform a certain action.

The counter-intuitive part is that it doesn’t take a lot of conversations for patterns to emerge.

It only takes 5 usability tests to uncover 80% of the problems. — Steve Krug

While a pattern based on a handful of conversations will not pass the statistical significance test, it does provide much stronger problem evidence strength than a postulated theory. Validate qualitatively, verify quantitatively.

Discovery is a key step to uncovering signals from the noise (insights).

Every validation campaign should aim to either reference some prior discovery activity or explicitly call out for some discovery activities as part of the campaign.