Technical SEO Audit Template

Pull live robots.txt; map every Disallow rule against actual indexable URL patterns; note any AI crawler (ChatGPT-User, OAI-SearchBot, PerplexityBot, ClaudeBot, GoogleOther) that's blocked.

Default Cloudflare hardening or default WordPress security plugins frequently disallow AI crawlers entirely.

Compare crawled indexable URLs against sitemap; note orphans (pages not in sitemap) and ghosts (sitemap entries that 404 or noindex).

Sitemap generated once at deploy and never regenerated; or auto-generation includes 404s and noindex pages.

Pull Indexed / Not indexed / Excluded; investigate every category over 100 URLs; tag root cause for each major bucket.

Unknown 'Crawled — currently not indexed' bucket usually traces to thin or duplicate content; 'Discovered — currently not indexed' usually means crawl-budget exhaustion from low-value URL clusters.

Spider site, export rel=canonical column, find non-self-referential canonicals; investigate every cross-domain canonical and every conflicting cross-page canonical.

Faceted nav or pagination canonicalising to root, fragmenting category-page rankings.

Crawl all internal links; flag 4xx and 5xx responses; verify all 301 redirect chains resolve in ≤2 hops.

Redirect chains 5+ hops deep after a CMS migration; intermittent 5xx during peak traffic.

View source vs rendered HTML on top 20 pages; verify H1, primary copy, and internal links appear in initial HTML.

JS-only rendered SPA pages where the H1 is empty in source HTML — invisible to AI crawlers, weakly indexed by Google.

If multi-locale: confirm bidirectional hreflang on every page in every locale; verify x-default; spot-check Quebec FR pages for hreflang reciprocation.

Unidirectional hreflang from EN to FR but not back — suppresses French SERP visibility.

Pull Search Console enhancements report; run top 10 page types through Rich Results Test; document every schema in use plus errors and warnings.

Critical schema (Article, LocalBusiness, FAQPage) missing on the page types where it would matter most.

Run PageSpeed Insights on top 10 pages; pull field data (CrUX) where available; test on mid-tier Android over Canadian 4G.

Hero image not preloaded, oversized hero on mobile, web font causing LCP delay.

Field data; manual interaction test on top revenue pages; flag any element with interaction latency over 200ms.

Heavy JS handlers on navigation, autocomplete, or filter UI.

Field data; manual scroll test; identify shifting elements (ads, images without dimensions, late-loading fonts).

Hero or above-fold images without explicit width/height; web font swap causing layout shift.

Curl with -w timings from a Canadian region; flag pages over 800ms TTFB.

Origin server in US East with no Canadian edge cache; database queries on uncached pages.

Audit top 10 pages for image size; flag any image over 500KB; verify WebP or AVIF where appropriate.

Multi-MB hero images, no responsive srcset, JPEG/PNG instead of modern formats.

Network panel on top 5 pages; flag any single JS bundle over 250KB gzipped.

Unsplit vendor bundles, full UI library shipped on every page, unused locale data from date libraries.

Inspect response headers on static assets; verify cache-control max-age and immutable flag where appropriate; verify HTML cache strategy (CDN edge, no client cache for HTML).

Static assets cached for one hour instead of one year; HTML cached aggressively at CDN, breaking authenticated experiences.

Spider site; export title column; sort by duplicates and length; flag missing, truncated, or duplicated titles on indexable pages.

Generic CMS-default titles ('Page – Site Name') on hundreds of pages; brand name leading every title at the expense of the actual query.

Same crawl; flag missing, truncated, or duplicated descriptions on top-traffic pages.

Auto-generated descriptions duplicating the page H1; or no description, letting Google synthesise.

One H1 per indexable page; H1 mirrors search intent; not the brand name on commercial pages.

Multiple H1s due to template confusion; H1 is brand or 'Welcome to'.

Spot-check top 20 pages for H1→H2→H3 logical nesting; flag skipped levels.

H1 → H4 with no H2/H3 due to design styling overriding semantics.

Crawl with internal-link-counts column; flag indexable pages with 0 or 1 inbound links.

30%+ of indexable pages effectively orphaned despite being in sitemap.

Audit top 50 images; flag empty, generic ('image1.jpg'), or stuffed alt attributes.

Bulk import migrations with empty alt on every image.

Pull top-10 ranking pages for 5 target queries; compare word count, heading count, FAQ presence, schema, and original elements.

Underbuilt pages competing with 2,000-word top-10 cohorts on 400 words of content.

Audit top 20 URL patterns; flag mixed case, underscores, parameter-heavy URLs in canonicals, excessive depth (>4 segments).

Tracking parameters on canonical URLs; CamelCase URLs from a CMS upgrade.

Verify site-wide on every indexable page; check stable @id; check that all other schema @id-references this Org node.

Multiple Organization nodes with different @ids fragmenting the entity.

Spot-check top 10 articles for headline, description, datePublished, dateModified, author (with @id), publisher, mainEntityOfPage.

Missing dateModified; author as plain string instead of Person @id.

Validate every page with FAQPage schema in Rich Results Test; verify Question/Answer text matches visible content; flag fabricated FAQs.

FAQ markup with answers that don't appear on the visible page (manual-action risk).

Verify provider @id, areaServed, priceCurrency CAD on Canadian-targeted offers, availability, itemCondition.

Missing priceCurrency, USD prices on a Canadian-targeted site, areaServed missing or overclaiming.

Verify schema BreadcrumbList exactly mirrors visible breadcrumb navigation.

Schema breadcrumb missing on key page templates, or includes hidden items not visible to users.

Trace every Review and AggregateRating schema to a verifiable third-party source; flag any self-reported review schema without source.

Synthesised AggregateRating with high stars and no traceable source — manual-action risk.

Pull referring domains from Ahrefs or SEMrush; flag toxic patterns (PBNs, comment spam, irrelevant TLD farms); review disavow file.

Inherited toxic links from a previous SEO vendor never disavowed.

Identify highest-equity referring URLs; verify the linked page is still live and on the right URL (not 301'd elsewhere).

Old high-value links pointing to redirected URLs that lost equity in a CMS migration.

Pull internal PageRank distribution; identify high-equity hub pages; flag commercially important pages that aren't receiving internal link equity.

Money pages (services, locations) buried in nav and starved of internal equity vs. blog posts that received heavy internal linking.

Audit anchor text distribution across internal and external links to top 10 commercial pages.

Generic anchors ('learn more', 'click here') on internal links; over-optimised commercial anchors on external links creating spam profile risk.

Verify rel=nofollow / sponsored / ugc applied appropriately; flag affiliate or paid links without disclosure.

Affiliate links treated as editorial; UGC sections (comments) without rel=ugc.

Crawl with broken-link checker; flag every internal 404; pull external broken-link list.

Hundreds of internal 404s linking to legacy URL patterns.

Verify ChatGPT-User, OAI-SearchBot, PerplexityBot, ClaudeBot, GoogleOther are not blocked at robots.txt or WAF level.

Cloudflare bot-fight mode blocking AI crawlers entirely; default robots.txt with overbroad disallow.

Run brand-name and key category queries in ChatGPT, Perplexity, Google AIO, Gemini, Claude; document who is cited alongside (or instead of) the brand.

Major competitors cited; brand absent or relegated to 'similar options' lists.

Verify every article points to the same Person @ids for author and reviewer; flag inconsistent or missing author attribution.

'Editorial team' as author, fragmenting author entity across the site.

Verify FAQ markup on the page types where buyers actually ask questions (services, products, pricing); flag eligible pages without FAQ markup.

FAQ content present but not marked up; or boilerplate FAQs across every page (low information gain).

Identify content that publishes original observations, data, or methodology; assess whether it's citation-friendly (clear methodology, named author, stable URL).

Pure synthesis content with no first-party data; or original research published without methodology section.

curl with realistic timeout (3s); test top 20 pages; flag any page that doesn't return full HTML within the AI crawler timeout.

Slow origin causing AI crawlers to time out before content loads, even though Googlebot succeeds.

Every finding tagged with severity (P0–P3) and effort (S/M/L); P0 + S findings are the immediate-fix list.

Audits delivered with 200 findings and no prioritisation; nothing ships.

Each P0/P1 finding includes: one-paragraph problem statement, exact URL pattern affected, recommended fix, validation steps, expected impact.

Findings written as marketing language; engineering team cannot translate to ticket scope.

Per finding, an explicit 'how we'll know it's fixed' test — URL inspection, header check, schema validator, etc.

Findings closed without verification, regressions go undetected.

Sequenced rollout plan tying P0 fixes to upcoming sprint cycles; dependencies surfaced.

Audit delivered as static document; no follow-through into sprint planning.

Define quarterly re-audit scope; identify drift-prone areas (sitemap completeness, schema validity, redirect chain depth).

One-shot audit with no re-audit; technical debt re-accumulates within 6 months.