More impressions.
Fewer clicks. What changed?
A falling click-through rate is a starting point for investigation. Work out which searches changed before deciding that every title needs a rewrite.
1. Make the comparison fair.
Use the same property, search type and filters for two complete periods. Record the dates. A 28-day comparison gives you the same mix of weekdays, but holidays, seasonal demand and site changes can still affect the result.
Check whether either period contains incomplete or delayed reporting. Keep site totals separate from the sum of a query table: detailed query reporting can omit some searches.
2. Check whether the query mix changed.
Consider a fictional page with these two groups of queries. The figures illustrate the arithmetic; they are not customer results.
| Query group | Previous: impressions / clicks | Current: impressions / clicks |
|---|---|---|
| Established relevant queries | 8,000 / 360 | 8,000 / 240 |
| Other queries | 2,000 / 40 | 7,000 / 60 |
| Total | 10,000 / 400 | 15,000 / 300 |
Total impressions rose 50%. Clicks fell 25%. Overall CTR dropped from 4% to 2%. But there are two different stories: new exposure added 20 clicks, while established queries lost 120.
Inspect the established query group first. Its CTR fell from 4.5% to 3% at the same impression count. The wider query mix explains part of the overall CTR decline, but not the loss of established clicks.
3. Find the affected pages and queries.
Open the page breakdown, identify the URLs losing clicks and filter to one page. Compare its query groups, then inspect device and country when those differences are relevant. Keep the filters visible in your notes.
Use average position as context. It summarizes reported appearances; a site-wide average can conceal changes in individual queries. A new query at a lower position can change that average even when an established query is stable.
If several pages appear for the same query, review what each page offers. Shared exposure alone does not prove that pages are competing harmfully.
4. Compare the result with the visitor’s need.
Read the affected page and its title. Does a visitor looking for a setup guide reach instructions, or a general product pitch? Are important details current and easy to find on a phone?
Check the search result context as well: layout, competing answers and intent may have changed. One search from your own device is a spot check, not a universal view of positions. Avoid assigning a cause solely from one screenshot.
Choose a specific hypothesis. For example: “People searching for setup steps land on an overview. Add accurate setup instructions and a descriptive title.” A useful improvement has a reason beyond increasing keyword repetition.
5. Make one clear change and record it.
Note the URL, date, query group, observed problem and intended improvement. Keep the original comparison. After Google revisits the page and enough new data accumulates, compare complete periods again with the same filters.
Separate what happened from why it happened. More clicks afterward is an observation; demand and other changes may also contribute. If the evidence is still sparse, keep collecting it instead of making another speculative rewrite.
theoretic.si’s Search Console insights bring comparisons and opportunity candidates into your workspace. Our weekly SEO review guide turns this investigation into a repeatable routine.
Metric definitions and report filters: Google’s Performance report documentation ↗.