Ruminating on... KPIs that are actually useful
We are starting Ruminating on..., the series where we think out loud about what we encounter while working with data, processes, and artificial intelligence.
In this episode, we ponder a somewhat uncomfortable question:
Are you using KPIs tailored to your business or the ones you saw in a template?
The problem is never measuring too little, it's measuring what belongs to others
Almost everyone measures something. The problem is that many dashboards are full of inherited indicators: the ones that came with the tool, the ones the previous company used, or the ones that appear in every industry article.
The result is usually a very complete panel, with a bunch of charts and numbers, that no one looks at because none of those figures change what you're going to do tomorrow.
And having more data doesn't necessarily mean better understanding what is happening.
A useful KPI answers a decision
We often use a fairly simple filter:
If this figure goes up or down, does anyone do anything differently?
If the answer is no, it's probably not a KPI.
That's why we try to work backwards: start with the decisions the business needs to make and then look for the data that helps make them.
Some ideas we try to apply:
- Start with the decision, not the available data. Just because a tool can measure something doesn't mean you need to measure it.
- Few and clear. A handful of indicators that everyone understands is better than thirty that no one can explain.
- With context. An isolated number, without a target, historical record, or comparison, tells you very little.
- Review them periodically. The business changes, the questions change, and the indicators should too.
It's the same idea we share in our handbook on how we work: measure to decide, not to fill out a report.
And it connects directly to what we discussed in Ruminating on... process automation: just as it's not a good idea to automate what you don't understand, neither is it a good idea to measure what you aren't going to use.
