Your first dashboard: five numbers worth tracking

First, count qualified arrivals: the people in the audience you are trying to serve. Define the source and avoid adding page views, sessions and people together as if they were interchangeable. A rise in visits from an irrelevant campaign can look like growth while producing no useful activity.

Second, measure activation: the share of those people who complete the first meaningful task. That might be submitting a usable enquiry, completing a booking or processing a document. Show the numerator and denominator beside the percentage. With a small audience, one extra completion can produce a large percentage change.

Third, look at return behaviour over a period that fits the product. A daily workflow and an annual service should not share the same retention window. Group users by when they first activated, then compare how many return to complete another meaningful task.

Do not assume every product should be used more often. For some services, quickly solving a problem is the intended result. Choose a repeat-use or completion measure that reflects customer value rather than encouraging unnecessary activity.

Fourth, measure whether the core journey works. Track successful completions alongside failures and unresolved outcomes. A fast page is helpful, but a booking that looks successful while the confirmation fails needs attention. Give each failure category an owner and a route to investigation.

Fifth, measure the value delivered relative to its cost. Depending on the business, that could mean completed paid work, time saved or cost per processed item. Make the calculation explicit. If staff still spend ten minutes checking each automated result, include that effort in any claimed saving.

For every metric, record its event source, time zone, exclusions, update frequency and owner. Treat a missing value differently from zero. If tracking stopped yesterday, a chart showing zero activity should not quietly imply that customers stopped using the product.

Review the dashboard on a regular cadence. Ask which change is real, what might explain it and which action follows. Annotate releases and campaigns so that comparisons have context. Add a new metric only when it answers a question the existing set cannot answer; otherwise the dashboard becomes another inbox to ignore.