Start by defining what changed
Before looking for a cause, write down the question precisely. Did visits fall yesterday, did search traffic rise over several days, or did one page receive an unusual number of views? Note the metric, the dates, the size of the change, and the dashboard where you spotted it. Sessions, users and page views describe different things; a rise in page views without a similar rise in users may simply mean people viewed more pages per visit.
Compare like with like. Check the same days of the week, allow for incomplete data from today, and look at a longer view to see whether the change is outside the site’s usual variation. For a small organization, a handful of visits can make a chart look dramatic. Record the baseline as well as the change so colleagues can understand why you decided it needed investigation.
Check whether the measurement is trustworthy
A sudden change in a dashboard can reflect a change in tracking rather than a change in visitors. Ask whether anyone recently edited the site, changed a consent banner, updated analytics settings, moved pages, or altered a tag manager. If traffic dropped across nearly every page and source at the same time, a measurement issue is worth checking early. Visit the site yourself and confirm that pages load and that your approved analytics setup is behaving as expected.
Look for signs of duplicate tracking, missing tags, internal staff visits, or automated requests. An unexplained spike with very short visits concentrated on one page may deserve closer inspection, but no single metric proves it is bot traffic. Keep a brief note of any tracking changes you find. If the data is incomplete, label it as such rather than treating a corrected chart as a complete record of what happened.
Find where the change is concentrated
Break the traffic down one dimension at a time: landing page, channel, referring site, device type, and broad location if your tools provide it. You are looking for a pattern, not collecting every available report. A drop confined to search visits on a few landing pages suggests a different investigation from a drop affecting all visits on mobile devices. A spike on one article may have a simpler explanation than a site-wide increase.
For example, imagine a community group sees a sharp rise in visits on Monday. The overall chart offers little explanation, but the landing-page report shows that most additional visits reached its volunteer guide. A channel report then shows an increase in referrals. Those observations give the team a focused next question: where were people linking to the guide? They do not, by themselves, prove that the new visitors found it useful.
Compare the pattern with what changed
Check your own activity log, however simple it is. Did you send an email, publish a resource, run an event, change a navigation link, or launch a campaign? For a traffic drop, check whether an important page was removed, redirected, made harder to find, or affected by a loading problem. If search traffic changed, review the affected pages and any search reporting available to you before assuming a particular cause.
Treat plausible explanations as hypotheses. An email sent just before a spike is a useful clue, especially if tagged campaign visits also rose, but visitors may have arrived through several routes. Likewise, a quiet week does not necessarily mean a website change caused a dip. Note what evidence supports each explanation, what contradicts it, and what you cannot verify. This keeps the investigation useful without forcing certainty the data cannot provide.
Decide what to do and keep a record
Match the response to the finding. If a key page is broken or tracking has stopped, fix it and mark the period affected. If a useful resource is attracting visitors, check that its information is current, its next steps are clear, and it works well on a phone. If a fall appears limited to a completed campaign, there may be nothing to repair. Avoid making major content or design decisions from one unusual day.
Finish with a short incident note: what changed, what you checked, the most likely explanation, any action taken, and when you will review the result. A simple shared log makes the next investigation faster because you can compare traffic patterns with past site edits and outreach. The aim is not to explain every movement in a chart. It is to notice meaningful changes, rule out misleading data, and act where your organization can make a difference.
