Reflections on startup life: Week 64

I had such a great response to the post from last week — thank you!

In the last week we’ve been really focussed on behaviour, metrics and measurement. I’ve taken the opportunity at a few social events to quiz some people about and how they are using it. There’s some challenges. Primarily the issue right now is that our user growth is starting to tail off — this relates directly to earlier comments I’ve made about press.

This isn’t a bad thing (eternal optimist!). I mentioned a couple of weeks back that while press coverage is great, it drive peaks in acquisition. A much better strategy is to optimise the troughs. A good, solid ever increasing user base is what’s needed — then peaks on top of that is the icing on the cake. The right time to optimise the troughs is now when we can see it. So we aren’t trying to drive more PR at the moment, we’re trying to work on the trough.

The big plan of course is “make it go viral”,


Fortunately, unlike the video, we have some ideas about what this will be. There’s a lot of different information and metrics out there on Viral Coefficients, but really for our purposes the simplest description is “every active user needs to bring along some friends”.

How many friends? If ~50% of our users are active, and ~50% inactive (actually 25% of our users never properly activate at the moment, 45% we aren’t quite sure what they do and 30% are very active, so I’m using 50:50 as a rough split between Active / Inactive), then we need to collect three new friends for each active user. This works out as follows — for two of those new users, one will end up active, one not (50:50 split). But there’s a chance that our existing user might go inactive too — if we pretend it’s 50:50 each week, then bringing on one additional new user (with the same 50:50 chance of activating) means that we will on average replace the existing user. So three invites per active user will replace our active user base AND provide some growth.

Of course, if we increase the retention rate, or increase the activation rate, those figures will improve (and we’ll grow faster). Actually the retention rate is much better than 50:50 week to week — the chance an active user goes inactive is very low once you’re past the first few weeks, and evidence is suggesting that retention slowly picks back up after a few weeks too (so some users let us archive links for a while then find new use cases, or develop trust to switch to using as a primary tool).

We’ve been implementing some “passive” viral mechanisms in the last few days, so now you can tweet and facebook links from in a way that brings users back through with a “call to action” for anonymous users to try us out. This should help increase the general word of mouth, but it’s passive in the sense that this type of messaging, while it adds utility to the user sending the message, is not really a direct recommendation (just because you see someone using a tool would you rush out and buy one, vs. them telling you how great a tool it is).

Our big goal this week is to experiment with a much more direct viral feature. We’ve spent a lot of time thinking about this, talking with some users and designing it, it will be really interesting next week to describe how it goes.


  • New features. Really excited that the API is now at a point where it delivers enough functionality for the first of the iPhone apps to move to testing.
  • School holidays are finally over. Much more peaceful at home.
  • Some great meetups and social events, enjoyed getting out and about.

Lessons Learnt

  • Discipline. Constantly re-learning this one, but when you see somewhere that needs to refactored, refactor it then and there. It’s so tempting to rush in a new feature, but you incur technical debt which hurts in the future. We are increasingly becoming disciplined here — it hurts now, but it pays off big time (and is starting to pay off already — for every feature that takes longer than you’d like to implement, there are others that are quicker because of it).

Goal this week

  • “Make it go viral” aka implement some direct incentives for you to tell your friends about