At some point in 2018, YCombinator put 'metrics' engraved into my head in a Startup School lecture. I didn't think much of it – my projects were growing a little bit after all and with a few thousand users were still so tiny that optimizing all that 'big startup stuff' didn't make sense for me. Oh, how foolish.

Metrics are super important for startups, that much has been known for a long time. If you run a tight ship with a ton of employees, marketing efforts and a lot of money to be lost, you better make sure that your DAUs, MAUs, Churns and Funnels are in order to make the most out of everybody's time, but why would I need that?

At some point, the growth that MentorCruise experienced wasn't good enough for me anymore. In some months there was even negative growth – more folks left the platform than signed up, and my revenue went down. That's heartbreaking to see, so I got behind those magical metrics to see what was going on.

Your north star decides what you work on

Setting up metrics doesn't really take a lot of time. For me, I looked up what some useful metrics are, built a little dashboard and some database queries and was done with it. I displayed trial mentorships, churn, funnels – all that good stuff. It's a straightforward process that helps you tap around less in the dark, and is often the source you should take for prioritization.

In my case, I saw with shock that – while my churn metric at 5% was somewhat reasonable – my trial churn was at 25%. If you aren't clear what that means: All that hard work I put into my product is only 75% effective (a lot less if you look at all other metrics, but you know...). I'm wasting energy and time and money, all because I didn't want to be like the big companies who take it all so serious.

In some way, it was a really embarassing wake-up call. I was knees-deep into brainstorm how to get more traffic to the website, was churning out content like crazy, thinking about referral schemes and SEO – when 1/4 mentees left their mentor after the first week.

If you calculate your metrics, you should pick a north star and concentrate on it, plain and simple. In my case this was active mentorships. In order to optimize it, I had to improve acquisition and reduce churn. This has transformed my backlog from nifty little features that I wanted, to pain points and difficulties that my customers faced.

Optimization for the tiny ones

I'm trying to keep advice here as general as possible, so I'm not going too much into the steps I took to bring down churn since then (spoiler: onboarding was a pretty big part). The next thing you should look at – and excuse this horrible marketing keyword – is your funnel.

Even if there is a whole science around it, a funnel is quite a simple thing. You pour things in at the top (called 'leads', but imagine it is everyone visiting your site or clicking 'sign up') and at the bottom comes out customers (or in other words $). A perfect funnel would turn every lead into a customer, but that's rarely realistic. Every step it takes from visiting your site to giving you money is an opportunity to leave the site (or funnel).

In my case, my funnel was long. This was painful for me (because I lost a lot of people), but also painful for customers, because they want to give you money, but have a long way to go to get there. I initially used Mixpanel to visualize this, later switched to Aplitude (friendlier pricing) and found that it takes a visitor on the site around 7 to 8 steps to even come close to give me money. At every step, some folks dropped out. I put effort into bringing people on the website, but what for?

If you are running a product, you should concentrate on the following (in this order).

  1. Define north star metric
  2. Visualize funnel
  3. Improve funnel leaks
  4. Work on improving north star (e.g. paying customers)

So, where are we today? Since doing this late last year, my trial churn went from 25% to around 18%. I shortened my acquisition funnel by combining multiple steps into one (e.g. you can now create a user while filling out an application), which improved the overall funnel throughput from something like 2.8% to 5%. That's almost double!

There's much more work to do, but with my metrics in hand I can test things out, see if they improve, and always know what to work on next.