Set Up Analytics You'll Actually Use to Make Decisions
Most analytics setups measure everything and inform nothing. Track the few numbers that change what you do next.
Useful analytics measures the few numbers that change a decision, and ignores the rest. Most setups do the opposite: they collect hundreds of events, fill dashboards nobody opens, and still cannot answer whether last month's work paid off.
A metric earns its place only if a change in it would change what you do.
Start from the decision, not the data
Before adding a single event, write down the decisions you make often. Which page to improve next. Which channel to fund. Whether onboarding works. Each decision needs one or two numbers behind it. That short list is your analytics spec. Everything else is noise you will pay to store and never read.
Track outcomes, not clicks
Pageviews and sessions feel like measurement. They rarely move a decision. Track the actions that map to value instead:
- Sign-ups that reach activation.
- Demo requests from organic search.
- Trials that convert to paid.
When an outcome dips, you know to act. When a pageview dips, you mostly shrug.
Make one report you read every week
A dashboard you check once a quarter is decoration. Build one weekly report with five lines at most: the outcomes above, each against last week. Five honest lines you read beat fifty you ignore.
Keep the setup honest
Check that events still fire after every release. A tracking tag that silently broke three weeks ago will quietly lie to you for months. We test the key events on each deploy, because an analytics number you cannot trust is worse than no number at all.