← Back to Guides · SEO
We Audited Our Own Search Console Data: 4 Page Fixes That Actually Moved Position
Short answer: Google Search Console will show you exactly which pages are close to ranking and losing clicks for a dumb, fixable reason — thin intro copy, a missing comparison table, a generic title tag. We pulled our own Performance report, found four pages with real impressions but no clicks (or a position stuck in the 60s despite demand), and fixed each one with a specific change tied to the query that was already finding the page. Below is the exact audit: what Search Console showed, what we changed, and the general method so you can run the same pass on your own site.
This is not a theoretical SEO checklist. It's the four actual fixes we shipped on agentickit.co this week, with the Search Console numbers that triggered each one. If you run a solo SaaS and publish your own docs, pricing, or comparison pages, the pattern below applies directly — you likely have two or three pages sitting at position 20-70 with impressions and zero clicks, and the fix is almost never "write more content," it's "answer the query in the first 100 words."
Why impressions-with-no-clicks is the highest-leverage signal in Search Console
Most SEO advice tells you to chase keyword volume. But a page that already gets impressions for a query is a page Google has already decided to show for that query — the ranking problem is solved, or close to it. What's left is a click or relevance problem: the title/meta don't match intent, the page doesn't answer the query fast enough, or thin content is holding position down. That's cheaper to fix than earning a new ranking from zero, and Search Console's Performance report (Pages or Queries, filtered to a date range) tells you exactly where to look: sort by impressions, then eyeball CTR and average position.
We looked at four rows from our own Performance report. Two had real impressions and non-trivial position but 0% CTR. One was a near-miss (position ~29) on a doc-seeking query. One was already converting well and just needed a copy check.
The four pages, what Search Console showed, and what we changed
| Page | Query | What GSC showed | What we changed |
|---|---|---|---|
| /skills | cursor skills | 47 impressions, 0 clicks, average position 66 | Tightened the H1 and opening paragraph to define "what is a Cursor skill" in plain language inside the first 100 words, and added a direct interlink to the ranked 25 Best Cursor Skills in 2026 listicle so both pages cross-reinforce the same query intent |
| /compare/cursor-rules-vs-skills | cursor rules skills docs | 13 impressions, average position 29.1 — the strongest near-miss in the set | Confirmed the rules/skills/commands comparison table sits in the very first section (it already did) and tightened the framing to cite Cursor's own docs explicitly, since "docs" in the query signals someone wants a reference answer, not an opinion piece |
| / | agentickit | Already position 1.1, 50% CTR on 8 impressions | Left the title and meta description as-is — they already lead with the brand term and a value prop ("61 skills, 47 commands, 46 agents"), which is the right shape for a branded query already converting at 50% |
| /docs | agentickit | 4 impressions, position 3.8, 0% CTR | Rewrote the title tag from a generic "Documentation & CLI Quickstart" to lead with the brand and a concrete benefit: "AgenticKit Docs: 46 Agents, Setup in 2 Minutes" |
Two patterns show up across all four rows. First, branded queries (rows 3 and 4) don't need more content — they need a title tag that gives someone who already knows your product a reason to click the SERP result instead of typing the URL directly. Second, non-branded queries with poor position (rows 1 and 2) are usually a relevance problem at the top of the page, not a missing-content problem — Google is already finding the page for the right query, it just isn't convinced fast enough that the page answers it.
The actual pattern for a page stuck at position 60-70 with real impressions
If you find a row like /skills → "cursor skills" → position 66, don't start by writing more sections. Start at the top of the page:
- Does the H1 contain the query, or a close variant? If someone searches "cursor skills" and your H1 is "The Cursor Skills Pillar," that's close enough. If it's a product name with no descriptive text, it isn't.
- Does the first paragraph answer the literal question in the query ("what is X") in plain language, before any product pitch? This is also what feeds a featured-snippet-style definition box, and it's what an LLM citing your page will quote.
- Is there a table of contents or jump nav so both users and crawlers see the page's scope immediately, instead of inferring it from scrolling?
- Does a closely-related page link to this one, and does this page link back? A single orphaned page rarely climbs from position 60; a pair of pages that interlink on the same topic (a pillar page and a ranked listicle, for example) reinforces relevance for both.
That's what we applied to /skills: the H1 and lead paragraph already defined "a Cursor skill is a SKILL.md file that teaches the agent how to do one job," and we added the missing cross-link to the newer ranked listicle so the two pages reinforce each other for the same query cluster.
The near-miss pattern: position ~20-30 on a doc-seeking query
/compare/cursor-rules-vs-skills at position 29.1 for "cursor rules skills docs" is a different shape of problem. Position 29 with only 13 impressions means Google already trusts the page enough to show it on page 3, for a query that explicitly includes the word "docs." That word is a strong signal: the searcher wants a structured reference, not a narrative opinion piece. The fix isn't "add more words," it's "surface the structured answer (a comparison table) immediately, and frame it as grounded in the vendor's own documentation" rather than purely original commentary. We verified the page already opens with a rules/skills/commands table and a direct link to Cursor's own docs on commands in the first section — the structural bones were already right, so the remaining lift is incremental: keep that table as the literal first thing after the lede, and don't bury it under scene-setting.
For context on how this distinction actually plays out for builders, the discussion in the Cursor community is still active and somewhat unsettled — see the forum threads on skills vs. commands, rules vs skills vs commands vs hooks, and what's the difference between skills and rules. People are asking the exact query we're targeting, in the exact words, which is a good sign the page's framing is aimed at a real and current confusion rather than a made-up keyword.
Branded queries: don't touch content, touch the SERP snippet
Rows 3 and 4 are both the query "agentickit" — pure branded search, someone who already knows the product by name. For /, 50% CTR on 8 impressions at position 1.1 means the existing title ("Claude Code Agents, Skills & Commands for SaaS | AgenticKit") and description are already converting over half of branded searchers. There's nothing to fix there; changing it carries more downside (losing the working combination) than upside.
/docs is the opposite case: position 3.8, 0% CTR. A branded searcher landing on a docs page already knows what AgenticKit is — what they don't know yet is why a generic "Documentation & CLI Quickstart" title is worth a click versus just navigating from the homepage nav. We changed the title to lead with the brand and a concrete, scannable benefit: "AgenticKit Docs: 46 Agents, Setup in 2 Minutes." The number and the time claim are both already true and already documented on the page itself (the CLI quickstart), so the new title isn't a new promise — it's the existing promise surfaced one layer earlier, in the exact text that actually renders in the Google SERP.
A checklist to run on your own Search Console data
- Export (or filter) Performance → Pages, sorted by impressions descending, over the last 90 days.
- Flag any page with impressions > 0 and CTR = 0%. These are relevance/snippet problems, not ranking problems.
- Flag any page with position between 15 and 40 and an increasing impressions trend. These are near-misses — usually a structural fix (table, FAQ, clearer H1), not more words.
- For branded queries specifically, check CTR separately from non-branded. A branded query with low CTR is a title-tag problem; a branded query with high CTR means leave it alone.
- For each flagged page, open it cold and time how long it takes to find the literal answer to the query that's driving impressions. If it's not in the first 100 words, that's the fix.
- Ship the fix, note the date, and re-check the same query in Search Console 2-3 weeks later — rankings for near-misses can move within days once the structural signal (clear H1, table near the top, interlinking) is in place, but attribute changes conservatively since Search Console data itself lags and position is noisy at low impression counts.
If you're running this audit inside Cursor or Claude Code, AgenticKit's on-page SEO skill encodes this exact checklist (title, H1, meta, schema) so an agent can apply it consistently across pages instead of you re-deriving the method each time, and the /seo-audit command runs it against a page you point at.
We re-ran this exact audit a few weeks later and found the same pattern on a page not covered here — see the second audit pass and the title-tag-first rule that came out of it.
FAQ
What does it mean when a page has impressions but zero clicks in Search Console? Google is showing your page for that query, but nobody is clicking the result. That's almost always a title tag, meta description, or SERP-snippet relevance problem — not a ranking problem, since the page is already being surfaced.
Should I rewrite a page that's already converting well on a branded query? No. If a page already has high CTR relative to its position (like 50% CTR at position ~1 in the audit above), changing the title or meta risks losing a working combination. Leave it and check back after any future change.
Is position 29 on 13 impressions worth optimizing? Yes — that's a near-miss, the highest-leverage band in Search Console. The page is already trusted enough to rank on page 3; a structural fix (a comparison table, a clearer framing of doc-seeking intent) is usually cheaper than building a new page from zero.
How long after a fix should I expect to see movement in Search Console? There's no fixed timeline, and Search Console data itself lags real-time ranking by a few days. Re-check the same query after 2-3 weeks rather than days, and don't attribute a single week's noise to one change.
Do I need a separate tool to run this audit, or is Search Console enough? Search Console's free Performance report is enough to run this exact audit — filter by page, sort by impressions, and check CTR and position per query. No other tool is required for the diagnosis step.
A later pass applied the same method to ecosystem-framing gaps on /skills, /commands, and two other pages, which turned out to be a variant of the same root cause: the title tag not matching what the searcher actually typed. A fourth pass later caught the same "docs"-seeking pattern on /compare/cursor-rules-vs-skills, the comparison page this very audit also touched. A fifth pass went further and found a case this checklist didn't yet cover: a page already on page 1 stuck at 0% CTR, which needs a snippet-copy fix instead of a relevance fix — see pass five. A sixth pass found that two of the rows hadn't moved at all because not enough time had passed since the fix shipped, plus a cannibalization fix and an internal-linking fix that needed no content rewrite — see pass six.
Sources
- Cursor forum, Skills vs. Commands vs. Rules (referenced for query-intent context, not quoted)
- Cursor forum, Rules vs Skills vs Commands vs Hooks (referenced, not quoted)
- Cursor forum, What is the difference between skills and rules? (referenced, not quoted)
- Cursor docs, Commands
- Google Search Console Performance data for agentickit.co, pulled 2026-10-28 (first-party, not published externally)
Written by Rakshit Yadav. Published 2026-10-28, last updated 2026-10-28.
Related Tools & Agents
Run this checklist automatically
The on-page SEO skill and /seo-audit command apply this exact method to any page you point at.
See the on-page SEO skill →