Des Traynor proposes 5 stages of a startup's dashboard's evolution. The right metrics can and should drive your decisions, so a dashboard can be essential to make your decision process quicker and smoother. The five stages are:
- One step above "select count(user_id)": when you just need some idea what's going on. Don't overdo it at this stage.
- Aggregation: when you want to spot patterns across groups of users. At this stage, it becomes interesting to calculate averages.
- Trends: display the averages over time on a graph so that you can more easily spot trends
- Insights and actions: mark up the graphs so that you can better attribute causes to changes in trends.
- Projections: when you have enough data that you can take guesses as to how changes in your activities will affect future changes.
It's an interesting post, but there is one problem: the focus is too much on vanity metrics. Not a single one of the metrics used in the examples are actionable metrics. Vanity metrics are dangerous for a couple of reasons:
- They give you a sense of being in control even when you're not.
- They can lead you to make the wrong decisions. Data is dangerous.
It can be useful to collect these types of metrics anyway, because gut feeling can sometimes uncover connections that data could not, but always be wary of wild guesses as to why something is trending one way or another. Metrics should be directly linked to hypotheses which you're testing, and directly lead to decisions which will increase your sales.
If you read this far, you should follow my RSS feed here.