Page click-drop recommendations find pages that received clearly fewer clicks over the latest 28 completed days than over the 28 days before them. The finding states what was observed: how many clicks the page lost, and which of its numbers changed. It does not state why.
To see these pages, open the site’s Recommendations report: press the lightbulb with a number in the top bar, or add the report to the site with Add report, From a template, Recommendations (see Dashboards & reports). The action a page appears under depends on what changed most:
| What changed most | Action in the report |
|---|---|
| The page’s average position worsened | Investigate pages whose rankings have fallen. |
| The page’s CTR fell | Improve the title and description of the ranking page. This is the same action as for low-CTR recommendations, so the instruction appears once with the affected queries and pages listed together. |
| The page’s impressions fell | Investigate pages with fewer search impressions. |
| Impressions and CTR both fell, and neither clearly dominates | Review pages losing both impressions and CTR. |
Each page entry shows the page path, the clicks lost and a short reason that starts with the change in clicks, for example “Clicks fell from 240 to 120” followed by what changed. View page opens the page comparison in the Pages report with both saved periods applied.
What to do with a finding
Treat a finding as the start of an investigation:
- Open View page and compare clicks, impressions, CTR and average position between the two periods.
- Check the page’s queries over the same dates in the query table to see which searches lost clicks.
- Look for changes you made around the start of the second period: a new title, moved or removed content, changed internal links, a redirect or a canonical change.
- Check whether the whole site fell in the same period (see the site’s totals). A site-wide fall points to something other than this page, such as seasonality, demand or a technical problem.
- Track the page after you act. The finding disappears by itself once a later completed calculation no longer shows a qualifying drop.
The data shows that clicks fell. It does not establish the cause, and the report does not promise that fixing the page will bring the clicks back. If one of the page’s queries has just started to rank lower, it also appears separately under position-drop recommendations.
The evidence intervals
The rule compares two consecutive intervals of 28 completed days each, using Google Search Console web search data. Both the site totals and the page data must have been fetched and marked final for all 56 days. The saved dates stay fixed when you change the report’s date picker, and opening a page from the list applies both saved intervals to the comparison.
Preliminary days are excluded. If newer final data in one dataset leaves a gap in the required coverage, the calculation waits for complete coverage instead of presenting an older result as current. An interval that ended more than seven days before the calculation date is also treated as stale.
A page with no row on a final day received no clicks that day, so it counts as zero. A day that has not finished syncing is different: it is not treated as zero traffic, and the calculation waits until it is available.
When a page qualifies
Every condition below must be met before a finding is saved.
| Condition | Current rule |
|---|---|
| Previous clicks | At least 50 clicks in the earlier interval, or 0.1% of the site’s clicks in that interval, whichever is larger. The proportional threshold is rounded up. |
| Clicks lost | At least 10 fewer clicks than in the earlier interval. |
| Relative decline | Clicks fell by at least 30% compared with the earlier interval. |
| Clicks noise limit | The loss is larger than 1.96 × √(clicks in both intervals combined). This is the variation expected from clicks being counted events. |
| Daily noise limit | The loss is also larger than 1.96 × √28 × the standard deviation of the daily differences. Each day is paired with the same weekday 28 days earlier, so ordinary weekday rhythm is not mistaken for a decline. |
A page that fails any condition produces no finding. Pages with fewer clicks than the threshold are never assessed, because their day-to-day variation is too large to read.
A page with any row of impossible values (negative numbers, or more clicks than impressions) in the 56 days is skipped rather than interpreted.
A page whose address changed is not a decline
When a page gets a new address (a redirect, a new URL slug, a site move), the old address loses its clicks and the new one gains them. Nothing was lost, so the rule does not report it. A page that would otherwise be a finding is treated as moved when all of these hold:
- It has practically disappeared: a tenth or less of its earlier clicks are left.
- Another page practically did not exist before: a tenth or less of its clicks in the later interval were there earlier.
- The new page gained 0.7 to 1.5 times what the old page lost.
- It happened in the same week: the old page’s clicks fell below half of its best week in the week, or within one week of the week, that the new page’s rose to half of its best week.
- One new page explains one old page. Of several old pages, the one that lost the most is matched first, and it takes the new page whose gain is closest to its loss.
The conditions are strict on purpose. A real drop must not be explained away because some other page happened to grow, so a page that only fell (and did not disappear), or a page that was already getting clicks, never counts as the place the clicks went. When a move is not recognised, the old address is reported as a decline as before, so check redirects before acting on a large drop.
How the change is described
The reason line and the grouping come from a decomposition of the observed click change into two parts, calculated from the midpoints of the two intervals:
- Impression change: the change in impressions multiplied by the average CTR of the two intervals.
- CTR change: the change in CTR multiplied by the average impressions of the two intervals.
The two parts add up exactly to the observed click change. They describe the numbers; they do not identify a cause. If the page has no impressions at all in the later interval, its earlier CTR is used internally so that the loss is assigned to impressions.
The page is described by its dominant change:
- Position, if its impression-weighted average position worsened by at least one position or 20% of its earlier position, whichever is larger.
- Otherwise CTR, if the CTR change accounts for at least 60% of the negative contributions.
- Otherwise impressions, if the impression change accounts for at least 60% of them.
- Otherwise both.
A worsening position is shown as context that goes with the decline. Average position can also change when the mix of searches that reach the page changes, so check the page’s queries and your recent changes before choosing a fix. CTR and position are unavailable for a period with no impressions and are shown as “—”.
Reading the numbers
The clicks lost figure is the observed difference between the two intervals. It is not a forecast, and it does not estimate how many clicks you would regain by acting.
Each finding’s priority is clicks lost × evidence strength. Evidence strength is a score from 0.5 to 1 based on how far the loss is beyond both noise limits and on how many clicks the page had before. It is not a probability of recovering clicks. Action groups and the pages within them are ordered by their highest individual priority. Priorities and lost clicks from different pages are not added together, and the number on the lightbulb counts distinct pages and queries, not clicks.
Why this recommendation? shows both intervals’ clicks, impressions, CTR and average position, and the click change by component.
Limits and special cases
- Changed addresses are recognised only when they look like a move. The rule does not read redirects or compare the words in addresses. It recognises a move from the numbers alone (see above), so a move that happens slowly, over several pages at once or with very different clicks can still show as a decline. Check redirects before acting on a large drop.
- Page addresses are shown as Google reports them. Different addresses are different pages.
- Web search only. Image, video, news and Discover traffic are not included.
- Google’s row limits. Search Console returns a limited number of rows per day, so on very large sites some page rows can be missing from the data. See Data & limits.
- Brand and non-brand are not separated. The rule looks at all clicks to the page.
Which findings appear
All active findings appear in the list. The report has no dismissal or snooze controls. If a page no longer qualifies in a later completed calculation, its finding becomes resolved and leaves the list. If a resolved page qualifies again, it returns.
Saved results
Calculations run after sync. The findings, rule version, threshold values and numerical evidence are saved. Opening the report reads the saved results, does not run the rule again and does not change Search Console data. A page’s saved result is kept while newer data is incomplete, and the report then shows a short note that some suggestions use older data.