Minutes 0–5: check the inputs
Choose a fixed review day and a consistent reporting period. Start by checking whether the data is complete. Note tracking changes, site outages, new campaigns, or an unusual mention. Compare similar periods where possible, and remember that a holiday or promotion can make two weeks very different.
Write down visits or sessions, the action you care about, and its count. Keep the metric names exact. Users, sessions, and page views describe different things and should not be swapped between reviews.
Minutes 5–10: examine the route in
Look at the sources that brought visitors and the pages where they entered. In GA4, the Traffic acquisition report describes the sources of new and returning sessions; it is useful for reviewing session-level acquisition. See Google’s Traffic acquisition documentation for the report’s definitions and dimensions.
Take the two or three sources that matter most to your goal. Record what happened after arrival, not just the volume. If a source changed sharply, check whether a tracking label, campaign, or landing page changed before you attribute the movement to your work.
Minutes 10–15: find one friction point
Choose one page or journey to inspect. If people arrive but do not complete a form, try the form yourself. If a campaign attracts the wrong audience, compare the promise in the message with what the landing page delivers.
Write an observation separately from an explanation. “The guide received 80 sessions and two signups” is an observation. “The signup offer may be unclear” is a hypothesis. Keeping them separate prevents a convenient story from turning into a supposed fact.
Use counts alongside percentages. Two signups from 80 sessions is 2.5%, but a few additional signups would move that rate substantially. Treat small samples as a prompt to learn more.
Minutes 15–20: make one decision
Finish with one action, an owner, and a review date. The action may be to keep collecting data. A weekly review does not require a weekly redesign.
- Repeat: keep a useful activity consistent.
- Revise: change one part of the message or journey.
- Pause: stop an activity whose effort is not justified.
- Investigate: resolve a measurement or usability question first.
Save a short note: “We observed __. We think __. Next we will __. We will review on __.” Read last week’s note before starting a new experiment, so you do not quietly change direction without learning from the previous one.
Your review is complete when it produces a clear next decision. You do not need to inspect every chart.