← All articles
SEO7 min read

Google Search Console with Claude AI: Find CTR and Ranking Opportunities Faster

Connect Search Console to Claude through MCP Ads to explain drops, uncover CTR upside, and prioritize SEO actions without spreadsheet work.

Search Console is one of the best sources of SEO truth and one of the easiest to underuse. Most teams export data monthly, spend an hour filtering it, and still end up prioritizing the wrong pages. MCP Ads makes it faster to ask direct questions about page decay, CTR weakness, and ranking opportunity.
WalkthroughSee it on a real account

One minute of Claude reading a live Google Ads account through MCP Ads, asking before it changes anything.

Spot low-CTR pages quickly
Explain drops by page or query
Turn raw data into a ranked action list

Why Search Console analysis gets delayed

Teams know the data is valuable, but the review gets postponed because it usually means exporting queries, clustering pages, and checking what deserves attention first.

That delay is expensive. Some of the fastest SEO wins are already visible in impressions and CTR — pages that rank but rarely get clicked, or positions that slipped from 8 to 14 without anyone noticing.

What Claude can surface well

Claude is strong at finding pages with high impressions and weak CTR, pages losing clicks month over month, and query clusters close enough to page one to deserve work now.

The useful output is not another spreadsheet. It is a ranked action list with page URLs, the data behind each recommendation, and specific suggestions for titles and meta descriptions where CTR is the issue.

  • Find pages with over 500 impressions but under 3% CTR
  • Which pages lost the most clicks compared to last month?
  • Show me queries where I rank position 5-15 with high impressions
  • Explain the traffic drop on /blog/example-page since April

A page-by-page workflow that works

The most reliable workflow is to start with the full site overview — total clicks, impressions, average position — then drill into pages that moved significantly in either direction.

For each page worth investigating, ask Claude to show the top queries for that URL and compare the current period to the previous one. This usually takes five minutes per page and surfaces the real story faster than filtering in the Search Console UI.

Combining Search Console with GA4

Search Console tells you about visibility and clicks. GA4 tells you what happened after the click. Together, they reveal pages that rank and get clicks but fail to convert — a different problem than low CTR and one that needs a landing page fix, not an SEO fix.

With both platforms connected through MCP Ads, Claude can answer cross-channel questions in the same thread: which pages are visible in search but converting poorly, and which ones need the title fixed versus the page fixed.

Worked prompt examples and what comes back

Search Console has a small tool surface compared to the ad platforms, which makes it easy to learn what each question maps to. Here are the prompts we see used most, the tool Claude reaches for, and the shape of the result.

"Find pages with lots of impressions but weak CTR." Claude calls gsc_search_analytics with the page dimension for the period you name, gets back clicks, impressions, CTR, and average position per URL, and then does the filtering you described — high impressions, low CTR — in the conversation. The output is a short ranked list of URLs with the numbers beside each, followed by a note on which ones sit in a position where a title change is likely to matter (roughly positions 3 to 10) versus the ones that are simply too far down the page for CTR to be the problem.

"Which pages lost the most clicks compared to last month?" This is gsc_compare_periods, with the two date ranges filled in and the page dimension. The response lists each page with clicks and impressions in both periods and the difference, and Claude sorts it by the size of the loss. The same tool with the query dimension answers "which queries did we lose."

"What queries send traffic to /pricing, and did any of them move?" Claude uses gsc_page_keywords for the URL, which returns the top queries driving clicks to that specific page with CTR and average position for each. Ask for the previous period alongside it and Claude runs the comparison so you can see whether the page slipped on one head term or bled across many long-tail queries — two very different fixes.

"Is this page even indexed?" When a page has zero impressions and you expected some, gsc_inspect_url returns the indexing status and crawl information for that exact URL. A surprising number of "SEO problems" turn out to be a noindex tag left over from staging, and this is the fastest way to rule it in or out.

"Show me clicks and impressions day by day for the last 28 days." That is gsc_date_trend, and it is the tool to use when someone says traffic "feels" down. A daily series shows whether the drop happened on one day (an algorithm update, a site change) or drifted over weeks (content decay, a competitor).

A step-by-step monthly Search Console review

The routine below is what we recommend running once a month, or every two weeks on sites that publish often. Each step is one prompt and each builds on the one before it, so keep it in a single conversation. It assumes the property is connected through MCP Ads; if you manage several sites, start with "list my Search Console properties," which calls gsc_list_sites.

  • Step 1 — Site trend: "Compare the last 28 days to the 28 before on clicks, impressions, CTR, and position." A single-line answer that frames everything else.
  • Step 2 — Biggest movers: "List the ten pages that gained and the ten that lost the most clicks between those periods." This becomes the investigation list.
  • Step 3 — Query view of the losers: "For each losing page, show the top queries and how their position changed." Claude iterates through gsc_page_keywords and summarizes the pattern.
  • Step 4 — CTR opportunities: "Find pages ranking between position 3 and 10 with impressions well above the site average and CTR well below it." These are the title and meta rewrite candidates.
  • Step 5 — Near-page-one queries: "Show queries in position 8 to 15 with high impressions." A refresh of the underlying page often moves these more cheaply than new content does.
  • Step 6 — Technical sanity: "Check index coverage and tell me if the error or excluded counts changed." This is gsc_index_coverage, and it catches the crawl problem before it becomes a traffic problem. Add "and confirm the sitemap is still submitted" to run gsc_sitemaps too.
  • Step 7 — Action list: "Rank everything into a prioritized list of pages to fix this month, with the reason and the suggested change for each." Ask Claude to draft the new titles for the CTR candidates in the same pass.

Turning findings into title and meta rewrites

CTR work is where Search Console pays off fastest, because a title change ships in minutes and shows up in the data within a couple of weeks. The trick is picking the right pages. A page in position 2 with a mediocre CTR is probably competing with a featured snippet or an ad block and a new title will not fix that. A page in position 5 to 8 with impressions well above its neighbours and a CTR below them is the one to work on.

Once Claude has the shortlist, ask it to draft two or three title options per page using the actual queries the page ranks for, which it already has from gsc_page_keywords. Then ask for a comparison in a month using gsc_compare_periods filtered to those URLs. If the site runs on WordPress and the WordPress integration is connected, Claude can apply the approved title and meta directly to the post rather than handing you a list to paste — but keep the approval step; the draft is the automated part, the decision is yours.

Limits and mistakes to avoid

Search Console is an honest dataset with a few sharp edges. These are the ones that most often turn a good prompt into a misleading answer.

  • Reading average position as a ranking. It is an average across every impression, including ones from queries where you barely appear. A page "at position 12" may be at position 3 for its main query and position 40 for a dozen others. Always look at the query breakdown before deciding a page dropped.
  • Comparing periods that straddle a known algorithm update or a site migration. The change is real, but the cause is not on your page.
  • Expecting complete query data. Search Console anonymizes some long-tail queries, so the query rows never add up to the page total. That gap is normal, not missing data.
  • Chasing CTR on brand terms. Brand queries already have a high CTR and little upside; filter them out before ranking opportunities, and ask Claude to do the same.
  • Ignoring the two-to-three-day data lag. Search Console data for the most recent days is partial, so end comparison ranges a few days before today.
  • Stopping at visibility. A page that gains clicks but not conversions is a landing-page problem, which is why the GA4 pairing matters.

Bottom line

If your team already has Search Console access, the real win is faster prioritization and clearer explanations of what to fix next. The data has always been there. The bottleneck was always the time it took to make it actionable — the same gap connecting Google Ads to Claude closes for paid performance.

FAQ

Can this help with title and meta rewriting?

Yes. Once Claude identifies pages with CTR upside, it can also propose title and meta angles for the pages worth testing first.

Is Search Console enough for full SEO analysis?

No. It is strongest when combined with GA4 and your knowledge of the page and business goal.

How far back does Search Console data go?

The API returns up to 16 months of data. Claude can compare any two date ranges within that window.

Can Claude break Search Console data down by country or device?

Yes. Ask for the breakdown and Claude uses gsc_country_breakdown or gsc_device_breakdown for the period you name. Device splits are especially useful for spotting a mobile-only CTR problem that an aggregate view hides.

Does this cover Bing as well as Google?

Not through Search Console itself, but MCP Ads has a separate Bing Webmaster integration that exposes the equivalent query and page data, so the same review can be run for both engines in one conversation.

Ready to connect your marketing stack?

Connect Google Ads, Meta Ads, GA4, and Search Console. Ask Claude anything about your accounts.