Recently, I finally realized that piling on new features day after day is not the most effective way to improve software. As a programmer, it’s tempting to just keep building… but this approach does not leave room for data-based improvement.
“Growth Hacking” is a (relatively) new term in startupland. I’ve been aware of the basic ideas for a while, but I’ve been so focused on building (coding) my software that I haven’t paid much attention. In retrospect, this was a big mistake. Growth Hacking is all about using data to justify development decisions, allowing you to spend valuable development time where it is most needed.
Case in point: what good is Streamified’s awesome PDF report generator (an enterprise feature) if most users aren’t making it far enough in the conversion-funnel to get even close to using it?
Choosing the Metric
Growth Hacking is, at it’s heart, a series of experiments. Like with any experiment, we need to decide upon the key metric to track, and then attempt to maximize that value – preferably while spending the least amount of money as possible.
To begin with, I was most interested in the average visit duration it Google Analytics for our web-app. Many people leave Facebook open in a tab for hours at a time; my hypothesis was that if we could achieve the same goal in our app, it was an indication that our users were using our app as their de-facto social portal.
I also keep an eye on events per visit (eg, clicking on a link) in order to make sure that these sessions are not “dead” sessions, where the user is completely inactive (walked away from the computer).
1. 100% increase: Redesign
The first major change was the most work, but also the most necessary. Our previous UI/UX was clunky. Bringing on a UX designer to redesign the entire look of the app got us from ~4 minutes to ~8 minutes average (Nov 2012 to Jan 2013).
This meant more than just making the design cleaner and simpler. For example, our new logo does a much better job of conveying what our product is than the previous one. This makes a more cohesive experience, where the user knows what to expect from the product before signing up.
2. 75% increase: Eliminating Distractions
Over the year I spent developing it, our product had become fragmented. We had a barely-working Android app, tons of hidden features in the webapp, etc. I ruthlessly pared down the offering, focusing on the essential question: what do we want users to be using? Even if this meant losing hundreds of daily users by getting rid of the fragmented pieces, it contributed to the goal of making sure that the users who DO use the site use it extensively.
Result: increase from ~8 minute to ~14 minute average session duration (Jan-Feb 2013).
3. 735% increase: Refining the Goal Funnel
Everything we’ve discussed so far has been preparation for the “real show.” Ironically, all of this prep work was far more work in terms of development, yet it is only from this last step that we saw an astronomical growth in the average session duration. This is where the social / human understanding comes into play, as opposed to just We went from ~14 minute sessions to ~1 hr 43 minute sessions:
Here are some of the things I did:
- Attempted to get users to a minimum of 5 stream subscriptions. This insight came from a conversation with a friend who works at YouTube, who explained that this is one of their key drivers for engagement.
- “Integrated” tutorials to guide users to important actions (eg, a tooltip showing a user how to add more streams when he doesn’t yet have 5 streams, per the above goal) instead of “prescriptive” tutorials (“read this guide”)
- Emailed users who averaged underneath 5 minutes per session to find out why the product didn’t work for them. For some, of course, it was simply not what they were looking for — but others had valuable insights.
- Automatically emailed new members a personal welcome email with some quick links to help them get started, effectively answering common problems and inviting them to speak directly with me.
- Created a demo video, along with other resources that appeared on the first login to guide the user though setup. Of course, many users won’t watch it all – in fact, I don’t think the value comes from the video itself, but from the presentation of the fact that there are professional materials to back-up the value of the product.
Through all this, the events per visit has also skyrocketed:
Next Steps: Improving Repeat Visits
“Average Session Length” is not our ultimate goal: it is, in fact, a prerequisite. With long session times and many events per visit I can safely conclude that the people using are app are using it for long, engaged periods.
Next, I’m interested in the number of repeat visits. It’s not much good if someone uses the app for 3 hours, but never comes back again. The number of repeat visitors per day has actually stayed relatively flat, which tells me that this is what is preventing us from showing better growth in overall active users. At the same time, I can see that there are a fair number of users who are true addicts (sign in every day). So, somewhere in there we’re loosing users who seem to like the app (use it a lot at first) but then forget about it.
My current ideas are pretty simple: to email users who have not used the service in a few weeks and see if I can get them to come back (eg, perhaps offer them a discount code for our Premium/Professional packages).
What I’m afraid of is that I don’t want to seem like a nag. So – do you have any suggestions? Any growth hackers out there?