Open Search Console’s performance report and a pile of queries have tens of thousands of impressions but single-digit clicks. That’s exactly what GSC CTR analysis is for: the page is already on the first page, users saw the title, but their fingers didn’t land.
Ignore the site-wide average CTR — that number has zero decision value. Bucket queries by average position, work out a baseline for each bucket, then pick out words below half their bucket’s baseline. You usually end up with a 20 to 50 word list. These words already have their ranking; what’s missing is the persuasiveness of the title and snippet. Fix those and you’ll see click changes within two or three weeks, at far lower cost than writing a new article.
Give every position band its own baseline
Brand words can hit 40% CTR, while a 2% long-tail Q&A word is normal. Averaging them together into 6.3% is just a feel-good number. The usable approach: bucket by average position, take each bucket’s median as the baseline, then compare individual queries against it.
| Average position | Reference CTR | Anomaly threshold | Priority action |
|---|---|---|---|
| 1–3 | 18%–35% | Below 9% | Check whether the title got rewritten by the search engine or squeezed by AI Overviews |
| 4–6 | 8%–15% | Below 4% | Rewrite the first 30 characters of the title; add numbers and concrete payoff |
| 7–10 | 3%–7% | Below 2% | Improve the description; add FAQ structured data to grab the snippet slot |
| 11–20 | 1%–3% | Below 0.8% | Improve ranking first; rewriting titles yields little here |
The ranges in the table are starting references. After 90 days of data, replace them with the medians of your own site’s position buckets and judgments get much more accurate. Industries differ a lot — tool sites’ first-page CTR is generally higher than news sites’.
Three export tables stack into an actionable opportunity list
Looking only at the query dimension loses landing-page info; looking only at the page dimension doesn’t tell you what users searched. Cross the two dimensions and you’ll know which page’s title to change.
- Switch the performance report to the “Queries” dimension, stretch the date range to 16 months, and export impressions, clicks, CTR, and average position.
- Switch to the “Pages” dimension and export another set to confirm which landing page each word hits is the page you want.
- Open the “Query + Page” combination filter and export the top 300 combinations by impressions — this is the real working sheet.
- Merge the three tables keyed by page URL; drop rows that don’t match, don’t fill by guessing.
- Set filters: impressions above 200, average position below 15, CTR below half of the bucket baseline.
Running this flow manually every week gets abandoned fast. Better to wire it into an automated report — we wrote the full data-pulling script in the GA4 plus Sheets weekly dashboard automation; hooking the query report in only needs one more API call and the sheet updates itself every Monday morning.
Low CTR splits into four diseases, each treated differently
Same low CTR, wildly different causes. Classify first, then act — otherwise you rewrite titles for ages while the real problem sits in the SERP’s presentation.
- Title lacks persuasion: ranking 4 to 8, every other result on the page carries numbers and years, only yours is a vague noun phrase.
- Intent mismatch: users search “how much”, your title talks about “what it is” — however high the position, no clicks.
- Eaten by SERP features: featured snippets, video carousels, or AI Overviews above you siphon impressions from the top three and pull natural CTR down.
- Title truncated: mobile shows only 30-something characters, the main word gets pushed to the back, users glance and don’t see the relevant bit, then scroll past.
The second and third types are easiest to confuse; the judgment method is crude but effective: search once manually in an incognito window and see what the results page actually looks like. The ranking-without-clicks article breaks down a finer judgment flow — pair it with the four-type classification here.
Six hard rules for rewriting titles and descriptions
- Put the main keyword in the first 30 characters — that’s all the width mobile displays.
- Numbers beat adjectives: swap “detailed guide” for “7-step checklist”, swap “comprehensive comparison” for “12 tested in practice”.
- Only add years to time-sensitive words — tool reviews and policy explainers suit them; methodology articles that add years need yearly maintenance.
- Keep descriptions under 150 characters; the main word and one concrete payoff must appear in the first 80.
- Change only the title or only the description at a time — changing both at once means you can’t tell which worked.
- After changing, push once with “Request Indexing” in Search Console and wait 3 to 7 days for effect before looking at data.
| Before | Problem | After | CTR change |
|---|---|---|---|
| Keyword tool recommendations | No numbers, no filtering reason | 9 keyword tools tested: 3 free ones are enough | 2.1% to 5.4% |
| How to do a content audit | Generic command, no payoff promise | How to do a content audit: 5 metrics cut 30% of dead pages | 1.8% to 4.0% |
| GA4 event setup explained | Reads like product docs, no conversational tone | GA4 key event setup: 6 actions become attributable conversions | 2.6% to 6.1% |
How to confirm it was the title that worked
Compare 28 days before the change against 28 days after, keeping only queries whose average position moved within ±0.5. Exclude words whose position moved — otherwise you can’t tell whether clicks rose because the title improved or because rankings went up. Also set aside strongly seasonal words; anything 11.11-related looks like it’s rising in October.
If the whole site’s CTR drops in sync rather than individual words, change direction. First check the Search Console coverage report to confirm whether a batch of pages fell out of the index — that’s a crawl and indexation-level problem that a hundred title rewrites can’t fix.
Five things to fit in before landing
Today, export 90 days of query data, split into four buckets by average position, and compute a median CTR baseline for each; filter words with impressions above 200 and CTR below half their baseline, take the top 30 by impressions descending as an opportunity table; tag each word with one of the four classifications and only start on “title lacks persuasion” and “title truncated” first; rewrite titles by the six rules, no more than 10 in a batch, recording change dates; 28 days later revisit the data, distill effective rewrite patterns into a template, and run the next batch. Run through these five and CTR genuinely becomes yours to keep.
FAQ
What CTR counts as normal in GSC?
There’s no universal standard — look by average position buckets: 1-3 around 18%-35%, 4-6 8%-15%, 7-10 3%-7%. It only counts as an anomaly when it’s below half of its bucket’s baseline.
How do I export GSC data?
First switch to the “Queries” dimension and pull 16 months, then export another under “Pages”, finally use the “Query + Page” combination to export the top 300 by impressions; merge the three by URL and filter.
What if there are impressions but no clicks?
Classify before acting: title lacks persuasion, intent mismatch, siphoned by featured snippets, title truncated. Rewriting titles only works for the first two; changing the last two is useless.
How do I rewrite titles effectively?
Main word in the first 30 characters, numbers over adjectives, description under 150 characters with the main word and payoff in the first 80. Change only the title or only the description — never both at once.
How do I confirm it was the title?
Compare 28 days before against 28 days after, keeping only queries whose average position moved within ±0.5. Exclude moved-position words and set seasonal words aside.
