Low-CTR recommendations identify non-brand queries that rank on the first page but receive fewer clicks than comparable queries on your site. To see each query, suggested action and conditional click estimate, 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 report groups findings by suggested action. Low-CTR queries appear together under Improve the title and description of the ranking page. Near-top-three recommendations, which assess queries in positions 8–20 against the site’s top-three CTR, have their own action group. A query can appear in both groups, with separate evidence and estimates for each action. When a page’s clicks have fallen mainly because its CTR fell, that page is listed in the same action group; see page click-drop recommendations.
The suggested action is to review the ranking page’s title and description against the query’s intent, then track CTR after making changes. A recommendation identifies a difference in the data; it does not establish why that difference exists or guarantee that editing the page will fix it.
The evidence interval
The rule uses 28 consecutive completed days of Google Search Console web search data. Both site totals and query data must have been fetched and marked final for every day. The saved interval does not change with the report’s date picker. Opening a query from the list applies its evidence dates to the query table.
Preliminary days are excluded. If newer final data leaves a gap in the required coverage, the calculation waits for complete coverage rather than presenting an older interval as current. A completed interval ending more than seven days before the calculation date is also treated as stale.
Existing findings can remain visible while coverage is stale. A short warning appears when a listed finding belongs to a rule that needs newer data. Findings update automatically after syncing brings in complete data. A site without enough completed history shows a coverage message instead of an estimate.
Your site supplies the benchmark
Brand queries are excluded from both opportunities and their benchmarks. The rule uses the site’s configured brand terms, or terms inferred from its domain when no terms have been configured.
Eligible queries have an impression-weighted average position from 1 to 10. That average is rounded to the nearest whole position for these six peer groups:
| Peer group | Rounded positions |
|---|---|
| 1 | 1 |
| 2 | 2 |
| 3 | 3 |
| 4–5 | 4 or 5 |
| 6–7 | 6 or 7 |
| 8–10 | 8, 9 or 10 |
The expected CTR comes from other non-brand queries on this site in the same group. The query being assessed is excluded from its own benchmark.
CTR is total clicks divided by total impressions. The peer CTR therefore gives more weight to queries with more impressions; it is not an average of their individual percentages.
When a query qualifies
Every condition below must be met before a recommendation is saved.
| Condition | Current rule |
|---|---|
| Query sample | At least 100 impressions, or 0.1% of the site’s total impressions in the interval, whichever is larger. The proportional threshold is rounded up. |
| Comparable sample | At least three other queries in the same position group. Each peer must meet the same impression threshold. |
| CTR difference | The query’s CTR is at least 30% below its peer CTR, measured relative to peer CTR. |
| Potential impact | Matching the peer CTR would produce at least five extra clicks in the evidence interval. |
| Sampling uncertainty | The upper end of the query’s 95% Wilson CTR interval is below the lower end of the peers’ interval. This avoids acting on a gap that could readily be explained by a small sample. |
For example, a query with 1,000 impressions and 10 clicks has a CTR of 1%. If its eligible peers have a CTR of 3%, the relative gap is about 67% and the conditional click difference is 20. It still needs enough peers and separation between the statistical intervals to qualify.
Queries without enough comparable peers produce no low-CTR finding. An empty list can mean that the available queries do not meet all the conditions; it does not mean that every page is performing as well as possible.
Reading the impact estimate
Potential extra clicks are calculated as:
Query impressions × (peer CTR − query CTR)
The figure applies to the same impressions and 28-day interval as the evidence. It answers what the click difference would be if the query matched its peers. It is an estimate, not a forecast of future traffic or the promised effect of editing a title.
Each query entry shows its CTR comparison and the conditional click estimate for that group’s action. Why this recommendation? reveals the query and peer sample sizes, each conditional estimate and its assumptions. Estimates from different actions are separate scenarios and are not added together.
Open View query to inspect that query in the table over the matching evidence dates. When findings for the same query have different saved intervals, separate links keep those dates distinct.
Each finding’s priority is estimated extra clicks × evidence strength. Action groups and the queries within them are ordered by their highest individual priority, without adding priorities together. Each group shows ten entries at first. Show more adds the next ten, and the button says how many it adds, at most ten (for example “Show 7 more queries”); in a group of pages it says pages, and in a group that mixes queries and pages it says items. The number on the lightbulb counts each affected query and page once, however many groups it is in, and a fall of the whole site once. Evidence strength is a conservative score from 0.5 to 1 based on sample size, peer diversity and the separation between the CTR intervals. It is not the probability that a change will gain clicks.
Which findings appear
All active findings appear in the list. The report has no dismissal or snooze controls, and an old dismissed or snoozed account state does not hide a finding. If a later completed calculation finds that the query no longer qualifies, the finding becomes resolved and leaves the list.
Saved results and data limits
Recommendations are calculated after sync and saved with their numerical evidence, rule version and threshold values. Opening the report reads those saved results; it does not run a new calculation or edit the underlying Search Console data.
The benchmark reflects the query rows that Google makes available. Google omits anonymized queries from the API, so query totals can be smaller than site totals. The rule cannot assess a query that is missing from the available data. See Google’s explanation of Search Console data filtering and Data & limits.