Full-site SEO audit

SEO Audit: addition-labs.com

Audit date: 2026-09-11 · Crawl: 10 pages (user cap) of 168 in the sitemap · Method: render_page.py, fetch_page.py, parse_html.py, sitemap_discovery.py, content_quality.py, analyze_visual.py, capture_screenshot.py via the claude-seo launcher; specialist checks run inline (no subagents available).

Executive Summary

SEO Health Score: 70 / 100

Business type detected: Agency. Signals: /case-studies (5 named clients), /industries (12 pages), client logos and "our partners" on the homepage, /services with four channels. A /pricing page is a secondary SaaS-like signal but the homepage copy ("the new-gen marketing agency for ecommerce and B2B brands") settles it. No local-service signals (the Sheridan, WY address in schema is a registered-agent address, no service area or map), so seo-local was not run. No e-commerce or drift baseline. Google API, Moz/Bing backlink and DataForSEO credentials are not configured, so those enrichments were skipped.

Top 5 issues

1. Navigation, footer and the homepage case-study rail are JavaScript-only. Raw HTML of the homepage has 5 <a> tags (4 service links + #contact); the rendered DOM has 216 (96 unique). <nav> does not exist in the server response. Every crawled sub-page shows the same pattern (3 to 13 raw links). Google renders JS so this is not an indexing block, but the internal link graph, the client proof ("10,000+ dog parents", "100,000+ people in Turkey") and 43 of the 53 discovered URLs are invisible to any fetcher that does not execute JS, which includes most AI-search fetchers. The four resource sections (guides, glossary, compare, tools) are unreachable from the raw HTML of any crawled page.
2. Hero images are 1 MB PNGs. seo.png 1,045,633 B, aeo.png 1,050,206 B, google-ads.png 1,129,315 B; the fourth panel (chatgpt-ads-official.webp) is 180 KB, which shows the target. No width/height, no srcset, no fetchpriority, and Cache-Control: max-age=0. seo.png is the visible hero panel and the likely LCP element.
3. Broken link in the rendered nav: /shopify/chatgpt-ads returns 404 and is not in the sitemap.
4. No security headers beyond HSTS. CSP, X-Content-Type-Options, X-Frame-Options, Referrer-Policy and Permissions-Policy are all absent.
5. Duplicate hosts/paths return 200: https://www.addition-labs.com/ and /index.html both serve the page instead of redirecting (the canonical tag points to the apex, so this is mitigated, not solved).

Top 5 quick wins

1. Re-export the three hero PNGs as WebP/AVIF at rendered size and add width/height (expected ~100-200 KB each, like the existing webp).
2. Fix or unlink /shopify/chatgpt-ads.
3. vercel.json: redirect www. to apex and /index.html to /; add X-Content-Type-Options: nosniff, Referrer-Policy, and frame-ancestors; long-lived cache for /img, /fonts, /logo (JS/CSS already immutable).
4. Homepage meta description is 50 characters ("We grow businesses with AI powered search systems."); the /industries/b2b (157) and /industries/ecommerce (154) descriptions are cut mid-sentence ("...from the problem they describe to."). Rewrite all three.
5. Add <lastmod> to the sitemap: 0 of 168 URLs carry it.

Technical SEO (72)

Crawlability. robots.txt is User-agent: * / Allow: / plus the sitemap line; nothing is blocked for any bot. sitemap.xml is a valid single urlset with 168 URLs; sitemap_index.xml, sitemap-index.xml, wp-sitemap.xml correctly 404. Every URL discovered in the crawl is in the sitemap. The rendered homepage links to four URLs not in the sitemap: /cookies, /privacy, /terms (fine) and /shopify/chatgpt-ads (404).

Indexability. All 10 crawled pages: HTTP 200, self-referencing canonical, no meta robots, no X-Robots-Tag, lang="en", viewport meta present. /services/seo/ (trailing slash) 308-redirects to the clean path. Unknown paths return a real 404.

JS dependence. This is the audit's main structural finding. render_page.py --mode auto classified the homepage as not-SPA (body copy is server-rendered, 1,686 characters of extracted text raw vs 3,203 rendered), so the core copy is fine. What is missing from the raw response is the chrome: <nav>, the footer link columns (services, industries, resources, tools, legal), and the case-study rail with client names, numbers and outbound links. Raw-vs-rendered vocabulary diff: 241 words exist only after render, and they are the navigation and proof vocabulary (Industries, Resources, Case Studies, Pricing, Contact, Tools, Glossary, the client metrics).

Security. HTTPS everywhere, http:// 308s to https://, HSTS max-age=63072000. No other security headers.

Server. Vercel edge, X-Vercel-Cache: HIT, TTFB 0.136-0.145 s across three samples, HTML 37.8 KB.

Content Quality (70)

E-E-A-T. The homepage does the experience work: five named clients with linked sites and concrete claims (rendered only, see above). Service pages have none of it: no named person, no credential, no date, no reviewer, no proof block, and zero outbound links on any of the nine sub-pages crawled. /about is in the sitemap but outside the 10-page cap, so team signals were not verified.

Thin content. Raw word counts: /services/seo 440, /services/aeo 429, /services/chatgpt-ads 431, /services/google-ads 404, /services/seo/audit 416, /services/seo/content 403, /services/seo/consulting 414, /industries/ecommerce 374, /industries/b2b 270. These are short for commercial landing pages, and the three SEO sub-pages share one template (same H2 count, same image count, same schema). content_quality.py flags /services/seo as low-density (0.174) and /industries/b2b as repetitive (33).

What is good. content_quality.py: homepage 96, /industries/b2b 95, service pages 76-78; filler 0 and AI-pattern 0 on every page. The copy is specific and in a real voice. The FAQ answers on /services/seo and /services/google-ads are direct one-paragraph answers.

Duplicate content. None found among the 10 pages beyond the www/index.html duplicates above.

Readability. Not scored numerically by the tooling; sentences are short and plain on every page read.

On-Page SEO (72)

Titles. All unique, 20-48 characters, | Addition suffix. SEO Audit | Addition (20) and SEO Content | Addition (22) leave room for a qualifier ("for Shopify stores").

Meta descriptions. Present on all 10. Homepage 50 chars (too short); two industry descriptions truncated mid-sentence; the rest are 114-148 characters and read well.

Headings. One H1 per page. Homepage H1 is correct ("Turning Search Visibility into Sales.", with an aria-label). On the seven service pages the H1 wraps two spans, an eyebrow label and the headline, with no separator: a text parser reads AEO / AI VisibilityBeing cited is not being recommended. The homepage has two one-word H3s ("You", "We") flagged by parse_html.

Internal linking. Raw HTML: home 4 internal links, /industries/b2b 3, /services/aeo 8, /services/seo 13. No contextual body links from service pages to the guides, glossary or compare pages that support them.

Schema & Structured Data (78)

Current. Homepage: Organization (legalName ADDITION LABS LLC, email, PostalAddress, sameAs LinkedIn, logo) + WebSite linked via publisher @id. Sub-pages: Service (serviceType, provider, areaServed, description, url) and BreadcrumbList on most; WebPage on /services/aeo and /services/chatgpt-ads; FAQPage on /services/seo and /services/google-ads. All JSON-LD blocks parse as valid.

Validation issues. None syntactic. Consistency: /services/seo and /services/google-ads have Service + FAQPage but no BreadcrumbList; /services/aeo and /services/chatgpt-ads have BreadcrumbList + WebPage but no Service. areaServed: "US" on the SEO service while the cited clients are Turkish and EU brands.

FAQPage (Info). Google retired FAQ rich results for all sites on 7 May 2026; the markup earns no SERP feature. Leave it, do not add more for SERP reasons, no AI-citation benefit is claimed.

Missing opportunities. dateModified on Service/WebPage; a Person per service owner once one is named; BreadcrumbList everywhere from one template.

Performance (55, unmeasured CWV)

What was not measured. pagespeed_check.py returned PSI rate limit exceeded (240 QPM / 25,000 QPD) because no Google API key is configured, and CrUX field data needs the same key. There are therefore no LCP, INP or CLS numbers in this audit; the score is an estimate from resource weights and should be replaced by a real measurement (Phase 4).

What was measured.

• TTFB 0.136-0.145 s; HTML 37.8 KB; JS bundle 20 KB (immutable cached); two variable fonts preloaded at 29 KB each.
• Stylesheet /assets/planPicker-i2NZ610t.css is 228 KB, the only CSS file, render-blocking in <head> on every page.
• Hero panels: three PNGs at 1.05-1.13 MB each plus one 180 KB WebP; no dimensions, no srcset, no fetchpriority; images cached max-age=0.
• 86 of 91 homepage <img> tags have no width/height; 54 are loading="lazy", so late-arriving undimensioned images are a CLS risk.
• Playwright: 0 console errors; desktop render 7.7 s, mobile render 3.0 s (wall-clock for a full render on this network, not a CWV metric).

Images (50)

• No <img> is missing an alt attribute (0 of 91 on home, 0 on every sub-page). 64 of 91 on home are explicit alt="", mostly repeated platform logos; the same google-c.svg is alt="Google" in one place and alt="" elsewhere. Sub-pages: /industries/b2b 26/26 empty, /industries/ecommerce 35/36, /services/aeo 40/53.
• The four hero panels have descriptive alt text.
• 0 <picture>, 0 srcset on the homepage: 1 MB PNGs are sent to mobile as-is.
• OG image: 1200×630 PNG, 66 KB, present on all pages.

AI Search Readiness (82)

• Crawler access: nothing blocked; GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended are all allowed by the wildcard rule.
• llms.txt: /llms.txt returns 200 with a site description, the legal entity and address, and 168 links, the same set as the sitemap, each with a one-line summary. /llms-full.txt also 200.
• Citability: body copy is server-rendered on every crawled page; FAQ answers are self-contained; Organization schema gives entity clarity.
• Gap: the proof that makes the agency citable (client names, numbers, links) is JS-only on the homepage, and no page carries a visible date or dateModified (render_page publication_date: null).

Visual

Screenshots at desktop, laptop, tablet, mobile in addition-labs.com-audit/screenshots/. Desktop: H1, sub-copy, the four service panels and the "Schedule a meeting" CTA are above the fold. Mobile: the cookie banner occupies roughly the lower 40% of the viewport with a large empty block between its copy and the two buttons, pushing the hero sub-copy and CTA behind/below it. analyze_visual.py reports horizontal_scroll: true on mobile (no obvious offender in the capture; needs a manual check).

Action Plan

Priority definitions: Critical blocks indexing or causes penalties; High significantly impacts rankings (1 week); Medium is an optimisation (1 month); Low is backlog. Nothing found is Critical: the site is crawlable, indexable and secure at transport level.

Phase 1: Critical fixes (Week 1)

Phase 2: High-impact improvements (Weeks 2-3)

Phase 3: Content & authority (Month 2)

Phase 4: Monitoring & iteration (Ongoing)

• Configure a Google API key (claude-seo run google_auth.py --setup) and re-run pagespeed_check.py for real LCP/INP/CLS; the Performance score above is provisional until then.
• Leading indicators without re-auditing: Search Console index coverage for the 168 sitemap URLs (watch "Discovered, currently not indexed" for the resources sections), and the raw-HTML anchor count on /.
• Re-run this audit after Phase 1 and compare: raw link count, hero weight, index coverage.

Artifacts written

addition-labs.com-audit/FULL-AUDIT-REPORT.md, ACTION-PLAN.md, audit-data.json, findings/{technical,content,on-page,schema,performance,geo,images,visual}.md, screenshots/addition-labs_com_{desktop,laptop,tablet,mobile}.png.

PDF report generated: addition-labs.com-audit/Google-SEO-Report-addition-labs.com-full.pdf (via google_report.py --type full; the Google-data sections inside it are empty because no Google API credentials are configured).

Limits of this audit

• 10 of 168 pages crawled (user cap): the four pillar service pages, three SEO sub-pages, two industry pillars and the homepage. Guides, glossary, compare pages, tools, case studies, /about, /pricing, /contact were not fetched; findings about them are inferred from the sitemap and rendered links only.
• No Core Web Vitals numbers (PSI rate-limited without a key; no CrUX key).
• No backlink, SERP or keyword data (no Moz/Bing/DataForSEO credentials; Common Crawl domain cache empty).
• No Google Search Console or GA4 data.