The only metric that matters
Josh Elman
3K13

How To Find OMTM using Google Analytics

Josh Elman does a great job of explaining a popular growth hacking concept: One metric that matters (OMTM).

What’s interesting is that OMTM is almost always is an action that user should take to keep using your product:

  1. Facebook: 7 friends within 10 days of signing up
  2. Dropbox: put at least one file in one folder on one device
  3. Twitter: visiting at least 7 times a month

Other examples may be found in “Growth Hackers Conference” or “Lean Analytics” book

This concept is really exciting, examples are even greater, but how do you find your OMTM? Is there a secret formula to find one?

Here is the 5 steps approach we use (we have borrowed from Goldratt’s “Theory of Constraints”):

  1. Current engagement level — typically, OMTM deals with engagement metrics, such as retention. The first step is to understand what is your current engagement level. The recently added “Cohort Analysis” report can help you with this: e.g. retention rates of a health tracking app are displayed below.
Cohort Analysis — Google Analytics

2. Hypotheses on retention levers — now brainstorm with your peers on the actions which significantly contribute to engagement: e.g. duration of ads impacts user’s engagement on video streaming website or the presence of card on file defines Order Ahead app’s usage metrics.

3. Test hypotheses — create appropriate segments in Google Analytics and compare retention rates with baseline metric (measured in step 1). In this example we see that Hypotheses #1 segment of users has 2x higher monthly retention rates.

Test hypotheses — Segmens in Google Analytics

4. Subordinate the whole system — if you are lucky enough to find the segment that has significantly higher retention rates than baseline, then you should subordinate the whole business (product: from onboarding to monetisation, marketing: from acquisition to value proposition & etc) to increase the fraction of users from such segment.

5. Start all over again :-)


If you are lazy enough to come up with a set of hypotheses in step 3, then you could force the computer to do this job for you. I suggest reading “Data Smart” book by John Foreman, Mailchimp’s Chief Data Scientist for description and examples of varios data analysis techniques which might help you come up with hypotheses & even test them for you.