The SEO Audit: What We Check and Why

An SEO audit paints the full picture: what you're doing right, what needs work, and what's quietly costing you rankings and AI-search citations.

An SEO audit has one job: tell you, plainly, what’s helping your site get found, what’s holding it back, and what to fix first. Think of it as a SWOT analysis for search: strengths to protect, weaknesses to fix, opportunities to chase, and threats to watch.

Search changed a lot since we first published this checklist. Google still matters most, but buyers now also ask ChatGPT, Gemini, Perplexity, and Google’s AI Overviews, and those engines pick a few sources to cite. So a modern audit checks two things: can you rank, and can you be quoted. Below is the audit we run for clients, in the order we run it.

1. Set up the instruments

You can’t audit what you can’t see. Before anything else, make sure these are connected and collecting data:

• Google Search Console: verify the domain property (not just one URL prefix), submit your XML sitemap, and check the Pages, Core Web Vitals, and Manual Actions reports.
• Bing Webmaster Tools: Bing’s index also feeds Microsoft Copilot and several other AI assistants. It takes five minutes to import from Search Console.
• Analytics (GA4 or equivalent): with conversions defined. Traffic without conversions is a vanity metric.
• A crawler: Screaming Frog, Sitebulb, Ahrefs, or Semrush. You need a full list of URLs, status codes, titles, and canonicals.

If you’re starting from zero, give the data two to four weeks before you draw conclusions.

2. Crawlability and indexation

If search engines can’t reach or index a page, nothing else matters.

1. robots.txt: make sure it isn’t blocking content you want found. Also check whether it, or your CDN, blocks AI crawlers like GPTBot, ClaudeBot, PerplexityBot, or Google-Extended. Some hosts and CDNs now turn those blocks on by default. That’s a business decision, not a default to inherit.
2. XML sitemap: it should list only live, indexable, canonical URLs, and be referenced in robots.txt.
3. Status codes: fix 404s that have links or traffic, collapse redirect chains to a single 301, and remove soft 404s.
4. Canonicals: every indexable page should canonicalize to itself; duplicates (http/https, www/non-www, trailing slashes, URL parameters) should point to one version.
5. Index coverage: compare the pages you want indexed to what Search Console reports as indexed. Big gaps usually mean thin, duplicate, or orphaned pages.
6. Rendering: if key content or links only appear after JavaScript runs, confirm Google sees them (URL Inspection → View crawled page). Many AI crawlers don’t run JavaScript at all, so server-rendered HTML is the safer bet.
7. Internal links and depth: important pages should be reachable within three clicks of the homepage, with descriptive anchor text. Pages with no internal links (orphans) rarely rank.

3. Performance and experience

• Core Web Vitals: LCP (loading), INP (responsiveness), and CLS (visual stability), measured on real users in Search Console, not just a lab test.
• Mobile: Google indexes the mobile version of your site. Check tap targets, font sizes, and anything hidden on mobile that matters on desktop.
• Security: HTTPS everywhere, no mixed content, and a current CMS and plugins.
• Usability: a useful 404 page, consistent navigation, and no intrusive pop-ups that cover the content people came for.

4. On-page fundamentals

For every page that matters (home, services, key landing pages, top posts):

• Title tag: unique, about 50–60 characters, primary keyword near the front, brand at the end.
• Meta description: unique, about 150–160 characters, written to earn the click. Duplicate or missing descriptions are one of the most common issues we find.
• One H1 that states what the page is about, followed by a logical H2/H3 structure.
• The first 100 words answer the core question or state the offer. Don’t make readers (or engines) dig.
• Images: descriptive file names and alt text; modern formats (WebP/AVIF) at sensible sizes.
• URLs: short, readable, lowercase, hyphenated.

5. Content quality and coverage

This is where most of the ranking upside lives.

• Search intent match: does each page give searchers what they actually want: a guide, a comparison, a price, a location?
• Thin and outdated content: pages with little substance, or advice that’s no longer true, drag down how engines judge the whole site. Improve them, merge them, or retire them with a 301.
• Cannibalization: two pages targeting the same query compete with each other. Pick one and consolidate.
• Gaps: topics your buyers search for that you don’t cover, especially comparison (“X vs Y”), cost, and “best for [industry]” questions.
• Experience and trust (E-E-A-T): named authors with real credentials, case studies with specifics, dates that show content is maintained, and clear contact information.

6. Structured data

Schema markup tells engines exactly what a page is: an organization, a service, an article, an FAQ, a product, a local business. Check that it’s present, valid (Google’s Rich Results Test and the Schema.org validator), and consistent with what’s visible on the page.

At minimum: Organization (or LocalBusiness) sitewide, BreadcrumbList, Article on posts, and FAQPage where you have real Q&A content. Structured data won’t rank a weak page, but it makes a strong one easier to understand and to cite.

7. Local signals (if you serve a region)

• A complete, verified Google Business Profile for each location, with categories, services, photos, and a steady stream of reviews.
• Consistent name, address, and phone across your site, directories, and maps.
• A location page per office with genuinely local content, not the same page with the city name swapped.

8. Links and authority

• Backlink profile: who links to you, to which pages, and whether any of it looks spammy.
• Lost links: high-value links pointing at pages that now 404. Redirect those URLs and recover the equity.
• Mentions: industry directories, associations, partner sites, and reputable “best of” lists. These matter for rankings, and they matter even more for AI engines, which lean on third-party sources when deciding who to recommend.
• Paid links: don’t buy links that pass ranking credit. Sponsored links should carry rel="sponsored".

9. AI-search visibility (AEO)

This section didn’t exist in our original checklist. Now it’s often where the biggest opportunity is. (For the full picture, see AEO: When AI answers the question, are you the answer?.)

1. Ask the engines. Run your 15–20 most important buyer questions through ChatGPT, Perplexity, Gemini, and Google AI Overviews. Record who gets mentioned and cited. That’s your baseline.
2. Answer-ready content. Pages that state a clear answer in the first sentence or two, use question-shaped headings, and include specifics (numbers, steps, prices, names) get quoted more often.
3. Entity clarity. Your company name, what you do, where, and for whom should be stated the same way across your site, schema, Google Business Profile, LinkedIn, and directories.
4. Machine-readable summaries. An /llms.txt file and a plain facts or “at a glance” page give AI assistants a clean source to work from.
5. Crawler access. Re-check step 2: an AI crawler that’s blocked can’t cite you.

10. Turn findings into a plan

An audit that ends as a 90-page PDF hasn’t done its job. We sort every finding into:

• Quick wins (under two hours, immediate impact): titles, meta descriptions, broken links, redirects, alt text, robots.txt fixes.
• Strategic work (this quarter): consolidating overlapping pages, building out thin service pages, FAQ content, industry pages, review generation, and link outreach.

Then we set specific, measurable goals, for example: “Increase organic leads from 8 to 15 a month within six months,” or “Be cited by at least two AI engines for ‘marketing agency for manufacturers in Southern California.’” Re-run the audit every quarter. Search doesn’t stand still, and neither do your competitors.

---

Want us to run this audit on your site? We’ll send back a prioritized list, quick wins first, and walk you through it on a 20-minute call. Start a conversation.