App Marketing Analytics - KPI? Good. A/B testing and predictive analysis? FORGETTABOUTIT
Madison Indies hosted a great presentation today at by Luke Doolittle (Manager of Business Analytics at Musicnotes.com). The presentation was on marketing analytics that was tailored for indie game developers but also relevant to anyone who might use metrics and data to measure their business. And if this isn't you and you're in business... uh... sorry but you suck.
But don't worry, it pointed out some stuff that I definitely suck at! There is really no shame in sucking at something. Especially if you're working to get better. Anyway...
Some of my quick takeaways from Luke's presentation:
There are 5 ways in which you can use data:
*Descriptive - normal everyday stats (averages, etc.)
*Exploratory - next step , hunting for relationships from descriptive, deeper
*Inferential - Hypothesis testing, measuring uncertainty, regression
*Predictive - current events to predict future events (machine learning)
*Causal - change in one variable results in change in another all things held constant (A/B testing)
The last two predictive and causal are very advanced and very challenging to get good results. Not worth chasing for anyone but the most advanced/well funded. Live and work in the first 3 (descriptive/exploratory/inferential)!
To make metrics work for your business:
* Start with your goal
* Understand your user base
* Create KPI's. 1 is good. More than 3 is bad.
* Revenue is a terrible KPI. The more granular the KPI the better (clearer indication of what's really going on) and if you can tie it to the Goal in step 1, that's ideal.
I look forward to taking some of these points and using them to improve the user experience (and the way we measure feedback) in our apps. Thanks for sharing the knowledge Luke!
