Most ecommerce setups track a lot and explain very little. Start with the questions people ask every week, then work backwards to the data needed to answer them.
Start with the questions, not the tools
Write down the recurring questions across acquisition, merchandising, conversion and retention. For each one, note what somebody would do differently if the answer changed.
Put the basics in the right order
A solid setup has four parts: reliable tracking, clean data, agreed definitions and a regular review where somebody decides what to do.
If a metric never changes a conversation or a piece of work, ask why it is in the dashboard.
1. Instrumentation
Use consistent event names and stable properties across the storefront. Write down what each event is for and test it whenever the relevant part of the site changes.
2. Data quality
Check for missing events, duplicates, consent effects and revenue differences. A polished dashboard cannot make bad inputs trustworthy.
3. Regular use
A report is only useful if somebody reviews it and acts on it. Give each report an owner, a schedule and a clear purpose.
Example: measuring product discovery
Suppose the team wants to know whether collection navigation needs work. Track how customers reach products, including search, filters, collection pages and dead ends. Compare the journey by device and landing page. The numbers should help you find where discovery breaks down, not simply show that people clicked something.
Give the setup an owner
Name who approves definitions, makes tracking changes, tests releases and reviews the reports. Without clear ownership, the setup starts drifting as soon as the next campaign or storefront release arrives.
Before you call the plan finished
- Match each business question to the data needed to answer it.
- Define events, properties and shared metric definitions.
- Validate collection against real customer journeys.
- Give every report an owner and a regular review date.
The goal is fewer reports you can trust, not more dashboards.
If the underlying tracking is unreliable, start with the Measurement Foundation package. If the data is sound but priorities are disputed, the Growth Opportunity Review is usually the better fit.
Worked example