A senior SEO audit used to take eight hours.
You pulled crawl data. You pulled GSC. You pulled Ahrefs. You took Lighthouse screenshots. Half a day was gone before any thought began. The work was data collection, not strategy.
We compressed it. Every YARD account now starts with a 135-point ai seo audit. It runs in under 10 minutes. The senior opens the report with breakfast.
This post shows the exact stack. The prompt. The 135 checks. A real output snippet. Steal it.
What "135-Point" Actually Means
The 135 checks split across six buckets. Each one is a real test the audit runs. Each one maps to a fix a senior SEO would otherwise flag by hand.
Technical (32 checks). Crawl, indexation, sitemaps, robots.txt, canonicals, hreflang, redirect chains, status codes, Core Web Vitals, mobile usability, structured data validity, HTTPS.
On-page (28 checks). Title tags, meta descriptions, H1 uniqueness, header hierarchy, internal link mesh, image alt text, schema completeness, URL structure, breadcrumb consistency.
Content (30 checks). Word count vs SERP, depth signals, keyword coverage, freshness, FAQ presence, citation quality, EEAT signals, named-entity coverage, cannibalisation.
Off-page (20 checks). Backlink profile, anchor text spread, lost links in 90 days, brand mentions, GMB consistency for local sites, toxic link risk.
AI search (25 checks). LLM visibility on ChatGPT, Perplexity, Claude, Gemini, and Bing. AEO signals. AI Overview presence. llms.txt config. Citation rate.
Decay watchlist. Pages losing more than 20% clicks in 90 days. Dropping queries. Lost positions. Quick wins live here.
Schema markup is a hard EEAT signal in this list. Google says structured data helps it grasp page content and surface rich results. (Source: Google Search Central, 2025 — developers.google.com/search/docs/appearance/structured-data)
Q: How does AI search even matter in a 2026 audit?
A: It is the fastest-growing surface. Gartner forecasts a 25% drop in classic search volume by 2026 as users shift to AI agents and chat. (Source: Gartner, 2024 — gartner.com) If your brand is invisible on ChatGPT and Perplexity, you are losing share before clicks even start.
The Stack That Runs It
Three layers. All text and APIs. No SaaS SEO platform.
Layer 1: Claude Code. The brain. Reads the checklist. Runs each check. Writes the report. Lives in the terminal. Costs only the monthly fee.
Layer 2: MCP connectors. Each plug is one line. Each one pulls live data, not stale exports.
- Google Search Console MCP — last 90 days of clicks, impressions, CTR, position.
- Ahrefs MCP — current DR, top backlinks, top organic pages.
- File System MCP — local crawl files, blog markdown, site config.
- Fetch MCP — sitemap.xml, robots.txt, page HTML, schema blocks.
Layer 3: The site itself. Public URLs are fetched and parsed. No third-party crawler for most checks. Sites over 5,000 pages get a Python + httpx helper script.
Why MCP at all? Anthropic's docs call it "an open standard for connecting AI assistants to the systems where data lives." (Source: Anthropic / modelcontextprotocol.io, 2025 — modelcontextprotocol.io) One prompt. Many sources. Live data.
Q: Why MCPs and not a SaaS crawler?
A: A SaaS tool can't see your GSC or your file system. MCPs can. One prompt pulls live GSC, the current backlink graph, the sitemap, and the file tree in parallel. The audit ships in minutes, not hours.
The Master Prompt (Steal This)
The whole audit runs from one prompt. Keep it tight. Keep it specific. Name your MCPs. Name your output shape.
Run a 135-point ai seo audit for {brand_url}.
Stack:
- GSC MCP: last 90 days of clicks, impressions, CTR, position.
- Ahrefs MCP: current DR, top 100 backlinks, top 50 organic pages.
- Fetch MCP: sitemap.xml, robots.txt, llms.txt, top 200 pages.
- File System MCP: /content/blog/ markdown, /public/schema/.
Output a Notion-ready markdown doc with:
1. Executive summary — 5 bullets max. What is broken. What is healthy.
2. Top 10 prioritised fixes — ranked by impact / effort.
3. Full 135-check table — pass / fail / warn, one-line reason.
4. Decay watchlist — pages losing >20% clicks in 90 days.
5. AI search visibility — does the brand rank on ChatGPT,
Perplexity, Claude, Gemini, Bing for the top 20 queries.
Self-check before output: every claim cites its data source.
No unsupported recommendations.
That single prompt produces the full doc. Runtime is 6 to 10 minutes.
A Sample Executive Summary
This is what lands in Notion. Real shape. Names redacted. Five bullets is the cap.
Executive Summary — brand.com (run 2026-04-22)
- Health: 92 pass, 31 warn, 12 fail. Decay flag on 17 pages.
- Big win: missing Article schema on /blog/* (108 URLs).
Ship in one PR. Est lift: +8% blog impressions in 30 days.
- Big risk: 14 title tags over 60 chars on top revenue pages.
Truncated in SERPs. Costs CTR every day.
- AI search: 4 of top 20 queries cite the brand on Perplexity.
Zero on ChatGPT. llms.txt missing. Add and re-test in 14 days.
- Backlink: -23 referring domains in 90 days. One toxic spike.
Disavow file already drafted in /audits/disavow-2026-04.txt.
The senior SEO reads that in two minutes. Then opens the top 10 fixes. Then ships.
Five Issues We Find Almost Every Time
We have run this audit on every YARD account. Five things show up nearly every time. If you are skipping the audit, fix these five first.
- Missing or invalid schema. Article schema absent on blogs. FAQ schema missing on FAQ pages. Product schema with broken price fields. Found in 9 of 10 audits.
- Title tags over 60 chars. Truncated in SERPs. Costs CTR. Easiest fix in the audit.
- Shallow internal link mesh. Pillars should link to 8–15 cluster pages. Most pillars link to 1–3.
- Content decay on 15–25% of blog URLs. GSC clicks down 20%+ in 90 days. The pages exist. They need a refresh, not a rewrite.
- No llms.txt file. As of 2026, Search Engine Land confirms llms.txt is treated as a real signal by AI crawlers. (Source: Search Engine Land, 2025 — searchengineland.com) We add one to every YARD site.
Q: How accurate is the audit when AI is doing the work?
A: Hard checks (status codes, schema parsing, meta tag length, canonicals) are 100% accurate. Judgment checks (depth, intent, gap) are good enough to triage. A senior SEO still reviews the report before any fix ships.
What It Looked Like in Practice
a luxury rehab brand is a luxury rehab group in the region. Four centres. Zero SEO foundation on day one. No structured data. No llms.txt. No GMB hygiene. No sitemap submitted.
The first audit found 42 distinct issues in under 10 minutes. We ranked them. We shipped the top 12 in month one. Then 18 more in month two. Then a content rebuild in month three.
Quick Facts: Brand SEO transformation (Dec 2025 → Mar 2026)
- SEO impressions: 25K → 98K per month (+290%).
- Average position: 12 → 5.1.
- GMB Calls: +783%.
- GMB Chat: +438%.
- Audit-to-fix cycle: 4 months end to end.
The audit-then-fix loop is what made this work. No 135-check starting list, and the team would have argued for weeks. The audit removed the argument. Speed of debate dropped. Speed of shipping rose.
How the AI Search Checks Work
The AI search bucket is the newest part of the 135. It is also the most useful in 2026.
For the brand's top 20 queries, the audit asks five questions per query.
- Does the brand surface in ChatGPT's answer?
- Does the brand surface in Perplexity's citation list?
- Does the brand surface in Claude's response?
- Does the brand surface in Gemini's answer?
- Does the brand surface in Google AI Overviews?
That is 100 data points per audit. The report aggregates per engine.
The structural fixes behind those wins are short.
- Valid llms.txt file at the root.
- FAQ schema on FAQ pages.
- Article schema with a quotable summary.
- Citation-ready content (specific stats, named sources).
- Definitional sentences ("X is Y that does Z").
an Indian kids' fashion brand (kids ethnic fashion) is the proof point. The AI search bucket was the most useful part of the audit for them. We ran a six-week catalogue rebuild. We shipped the LLM SEO foundation on top. The brand was invisible on AI search at the start. It now ranks on ChatGPT, Perplexity, Claude, Gemini, and Bing. The wins span every primary category. Google impressions doubled. They went from 1.1M to 2.3M per month in the same window.
For a sister brand (sister brand to a luxury rehab brand), the same playbook took Google impressions from 3K to 15.3K per month. A 5x lift in one quarter.
For the resort (niche resort), impressions went from 10K to 64K per month. Direct room nights doubled from 795 to 1,600.
What Changes When 8 Hours Becomes 10 Minutes
The win is not just speed. It is cadence, role, and room tone.
It changes the cadence. Quarterly becomes monthly. Monthly becomes weekly for high-velocity content sites. Annual becomes part of every onboarding.
It changes the role. Juniors used to spend a day on data. Now they spend an hour on choices. Same outcome. Different work.
It changes the room. With the audit done, the team's time goes to fixing. We have shipped 31 of 47 surfaced fixes inside 30 days. That is the gain.
The honest limit: the audit catches things. It does not fix them. Most fixes still ship by hand. Content rewrites. Schema deploys. Link mesh updates. GMB tune-ups. We use Claude Code to deploy 10,000 SKUs of Product schema in one session. That lifted organic traffic 14% in one quarter. But the average account ships fixes the old way.
A senior writes the spec. A dev or writer ships it. The audit removes the bottleneck on what to fix. It does not remove the work to do it. That is fine. The fix work is the high-judgment part anyway.
If you want to try this yourself, start small. You do not need all 135 checks on day one. Start with three.
- Decay pass. Pull GSC clicks for the last 90 days vs prior 90. Flag pages down 20%+. Easiest decay pass on earth.
- Title tag pass. Crawl the sitemap. List title tags over 60 chars. Truncate kills CTR.
- Schema pass. Hit the top 10 traffic pages. Validate any
<script type="application/ld+json">blocks. Use the Rich Results test.
Three checks. One hour. You will find 8 fixable issues on a typical brand site. Add 10 checks per week. By month two you have a real subset of the 135 adapted to your stack. Teams that audit weekly outperform teams that audit yearly. That gap compounds. The math is brutal. The good teams ship faster than the slow ones can plan.
Want Us to Run This for You?
YARD is an AI-first growth marketing agency. We run many campaigns across multiple brands at and The 135-point ai seo audit is one of five repeatable plays we run on every account.
Take a recent case as the proof. We started with a 2-collection website. We rebuilt the catalogue. We shipped the LLM SEO foundation. We ran the audit on a weekly cadence. Six months later: Google impressions doubled to 2.3M per month. The brand now ranks on ChatGPT, Perplexity, Claude, Gemini, and Bing for category queries. That is gold for an LLM SEO post.
If your SEO program runs on instinct, we can fix that. We will run a v1 audit on your site in under a week. You get the prompt, the report, and the prioritised fix list.
We do not just hand you a doc. We help ship the top 10 fixes too. We pair the audit with our content, schema, and link teams. Most clients see decay reverse in the first month. Most clients see AI search visibility move inside 14 days of llms.txt going live.
The bottom line is short. The 135-point audit collapses a day of grunt work into 10 minutes. The seniority you used to need to run it is now needed to act on it.
Book a call. See how we'd run this for your site. Talk to YARD at yardagency.ai and we will start with a free first audit.
FAQ
Q: What does a 135-point SEO audit cover?
A: Six buckets. Technical, on-page, content, off-page, AI search, and a decay watchlist. Each bucket holds 18 to 32 checks. Together they map every fix a senior SEO would flag.
Q: Can AI replace a senior SEO consultant?
A: For data pulls, yes. For calls on what to ship first, no. The audit cuts eight hours of grunt work to ten minutes. A human still picks the top five fixes.
Q: What tools run the audit?
A: Claude Code is the brain. MCP connectors plug it into GSC, Ahrefs, the file system, and fetch. The site itself is read by HTTP. No SaaS crawler is needed.
Q: How accurate is an AI SEO audit?
A: Hard checks like status codes, schema parsing, and meta tags hit 100%. Judgment checks like depth and intent are good for triage. A senior SEO still signs off before any fix ships.
Q: Can my team run this in-house?
A: Yes. The 135 checks are public. The prompt is reusable. The hard part is picking which five fixes ship first. Most teams reach that skill by month two.
Q: How often should we run this audit?
A: Quarterly is the floor. Monthly works for fast content sites. Weekly mini-audits catch decay early. Every YARD account starts with one in week one.
Q: Why use MCPs over a SaaS crawler?
A: MCPs let one prompt pull live GSC data, the current backlink graph, and the sitemap in parallel. A SaaS crawler can't see your GSC. The data layer is open and yours.
Insights from Our Experts
Explore our latest articles on digital marketing strategies.
