Kaizen AI Lab — SEO + GEO Audit

76/100

Rating: Good

Note: provisional website-readiness score — judgment-weighted rubric against crawl evidence (table below), not an industry-standard tool score. Rankings, backlinks, Search Console, Core Web Vitals, and live AI-answer data were unavailable this run (see Limitations).

Strong foundation, targeted fixes needed. The site is technically healthy: all 101 sitemap pages resolve, redirects are consolidated correctly, structured data coverage is broad, and the GEO groundwork (llms.txt, AI-crawler policy, FAQ schema) is ahead of most of the market. The score is held back by six broken internal link destinations (13 link occurrences across 8 pages, including a dead booking CTA on three industry pages), an orphaned services page, a sitemap hygiene error, and a sitewide pattern of over-length titles and descriptions.

Category	Weight	Score	Notes
Technical foundation & indexation	25	20	All pages 200, clean redirects, HSTS/TLS 1.3; deductions for broken links, orphan page, noindex URL in sitemap
On-page optimization	20	14	Zero duplicate titles/descriptions; 85 titles and 35 descriptions over recommended length
Content depth & architecture	15	13	87 articles + 4 pillar guides + FAQ hub; internal-link gaps to money pages
Structured data & entity	15	12	Broad JSON-LD coverage; gaps on /contact/ and /industries/; brand-name inconsistency
GEO / AI-answer readiness	15	12	llms.txt live, AI search crawlers allowed; no live AI-visibility measurement yet
Authority & off-site signals	10	5	Partially unverifiable this run (backlink API access blocked — see Limitations)

Executive summary

✅ Three wins available now

Fix the dead /book CTA linked from 3 industry pages — a conversion path is currently 404ing.
Redirect 4 legacy pillar URLs to their live /guides/ equivalents — reclaims link equity and fixes 6 broken in-content links.
Link /services/ into the site — the primary commercial page has zero internal links pointing to it.

⚠️ Three biggest gaps

No measurement loop: no baseline for rankings, backlinks, or AI-answer visibility (API access blocked; no prior audit artifact found).
Entity ambiguity: the site alternates between "Kaizen AI Lab" and "Kaizen AI Labs" — this weakens knowledge-graph and AI-citation consistency.
Sitewide title/description length pattern suppressing SERP click-through on commercial pages.

The single most important action: fix the broken internal links (the /book CTA and the pillar-to-guides redirects). They are live revenue-path and equity leaks, each is a sub-hour fix, and both are fully in our control.

Improvements since the last audit

Honest status: no prior audit artifact could be located in the audit archive, session history, or connected storage, so a defensible improved/unchanged/declined comparison is not possible this run. This report is the dated baseline — the next audit will diff against it automatically.

Worth noting: the site already carries mature assets that are usually the output of prior optimization work — a populated llms.txt, an explicit AI-crawler policy in robots.txt, schema on 95 of 101 pages, and clean canonical redirects. We cannot attribute these to changes "since the last audit" without a dated prior report. If a previous audit exists in Drive, Notion, or email, send it over and a comparison appendix will be added without re-crawling.

Errors and hiccups found on the site

High Broken internal links (6 destinations, 13 link occurrences, all verified 404)

Broken URL	Linked from	Recommended fix
`/book`	/industries/hr-recruiting/, /industries/legal/, /industries/real-estate/ (booking CTA)	Point CTA to /contact/ or the Cal.com booking link — confirm intended target, then 301 `/book` as well
`/blog/pillar-ai-compliance-mid-market-guide`	/author/don-ho/ + 3 guide pages	301 → /guides/ai-compliance-mid-market/ and update in-content links
`/blog/pillar-ai-financial-services-compliance`	/author/don-ho/	301 → /guides/ai-financial-services/
`/blog/pillar-ai-governance-framework-guide`	/author/don-ho/ + 2 guide pages	301 → /guides/ai-governance-framework/
`/blog/pillar-ai-legal-ethics-risk-disclosure`	/author/don-ho/	301 → /guides/ai-legal-ethics-risk/
`/blog/ai-compliance-readiness-framework`	/blog/saaspocalypse-legal-tech/ (in-article link)	Repoint to the intended live piece (likely /guides/ai-compliance-mid-market/ or /blog/ai-compliance-stack/) — content owner to confirm

Pattern: the guide content was migrated from /blog/pillar-* slugs to /guides/* without redirects or a link sweep. Add the 301s (old URLs may hold external links or indexed history), then update the in-content links so users never bounce through a redirect. Note: /cdn-cgi/l/email-protection also appeared as a 404 in raw crawling, but that is Cloudflare's email-obfuscation endpoint working as designed in a non-JS crawler — it is not a site error and was excluded.

High /services/ is an orphan page

/services/ is in the sitemap but zero crawled pages link to it. The homepage and nav use anchor links (/#kaizen-services, /#adjacent-services) instead of the standalone page. Verified twice against raw HTML of home, about, industries, contact, FAQ, and blog. (Caveat: this check reads static HTML and cannot see JavaScript-injected navigation; worth one manual confirm in a browser, though the anchor-link pattern in the raw markup makes a JS-only /services/ link unlikely.) A commercial page with no internal links gets minimal crawl priority and no PageRank flow. Either link it from primary navigation and relevant guides/industry pages, or consolidate it into the homepage sections and remove it from the sitemap. Recommendation: keep and link it — a dedicated services URL is a better citation and landing target for both Google and AI engines.

High Sitemap hygiene: noindex page listed

/hero-preview/ is marked noindex, nofollow (correctly — it is an internal hero-video test page) but is included in the XML sitemap. That sends Google contradictory signals. Remove it from the sitemap; keep the noindex.

Medium On-page length patterns (template-level fix)

85 of 101 titles exceed 60 characters — largely a blog-title template pattern ("Long editorial headline. Second sentence. | Kaizen AI Lab", worst case 166 chars). Not a ranking penalty, but Google truncates or rewrites them. Prioritize commercial pages (about, guides, industries, web-labs) and the highest-impression articles; leave working editorial titles alone until Search Console CTR data says otherwise.
35 meta descriptions exceed 160 characters (up to 257). Treat as a snippet/CTR opportunity: rewrite the commercial pages first.
Homepage has 2 H1s ("Intelligence Engineered. Deployed." / "Marketing Reimagined. Omnipresent."). This looks like a deliberate two-brand design (Kaizen + Adjacent Media), not sloppiness — but one H1 with the second demoted to H2 gives search and AI engines a single clear topic statement. Flagged for a design decision, not an unqualified error.
Homepage title/meta vs. rest of site: the homepage brands as "Kaizen AI Labs — Intelligence, evolved" and describes "two companies in one," while interior pages brand as "Kaizen AI Lab" focused on compliant AI consulting. See entity consistency below.

Medium Structured data gaps

Six pages carry no JSON-LD: /contact/, /industries/, /accessibility/, /privacy/, /terms/, /hero-preview/. Priorities: add ContactPage + Organization/ContactPoint schema to /contact/ and CollectionPage/ItemList to /industries/. Legal pages and the noindex preview are low priority.

For awareness — not errors

The 4 decorative images on /contact/ flagged by automated tooling use alt="" + aria-hidden="true" — that is correct accessibility treatment, no action needed.
All three external conversion endpoints verified live: ACRA (acra.kaizenailab.com), Cal.com booking, Calendly.
No security.txt; optional, low priority.

What is working well

Indexation and delivery: 101/101 sitemap URLs return HTTP 200; http→https and www→non-www 301s are correct; canonicals present on 100/101 pages; zero duplicate titles or descriptions across the site.
Speed proxy: median HTML response 71 ms, max 118 ms (Cloudflare edge) — excellent server-side performance. (Full Core Web Vitals were not measurable this run — see Limitations.)
Security posture: TLS 1.3, HSTS with preload, X-Frame-Options, X-Content-Type-Options, Referrer-Policy, Permissions-Policy.
Structured data footprint: Organization (94 pages), Person (92), Article (87), WebPage (88), FAQPage, Service, ProfessionalService, OfferCatalog, Speakable — far above typical for a consulting site.
Content moat: a large editorial library (87 pages carrying Article schema across the blog and guides), 4 pillar guides, a deep FAQ hub (5,800+ words), and 6 industry pages — a real asset for both search and AI answers.

GEO (generative engine optimization) analysis

Already in place — ahead of the market

llms.txt is live and well-formed: positioning, founder credentials, services with pricing ($2,500 audit / $12,000+ implementation), tagline. This is exactly the extractable-fact format AI engines lift from. (Caveat: llms.txt is a voluntary, emerging convention — useful groundwork, not a guaranteed citation mechanism.)
robots.txt AI policy is deliberate and mostly coherent: AI search/answer crawlers and access tokens (OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, Claude-SearchBot, plus Applebot-Extended — technically Apple's generative-AI control token) are allowed; known training crawlers (CCBot, Google-Extended, FacebookBot, Bytespider) are blocked, with a Content-Signal directive. Two mismatches: GPTBot (OpenAI's model-training crawler) and ClaudeBot (Anthropic's general/training crawler, distinct from Claude-SearchBot) are both currently in the allowed group — allowing them contradicts the stated search-yes/training-no policy. Flag for Don: either accept OpenAI/Anthropic training use, or move GPTBot and ClaudeBot to the blocked group.
FAQPage schema + a dense Q&A hub — well-structured Q&A content is significantly easier for answer engines to extract and cite.

GEO gaps to close

Entity consistency (the big one): the brand appears as both "Kaizen AI Lab" and "Kaizen AI Labs" across titles, schema, and copy — confirmed in the Organization schema itself, which returns "Kaizen AI Lab" while the homepage title reads "Kaizen AI Labs." Inconsistent entity facts make it harder for search and AI engines to reconcile the brand into one entity, which weakens citation and recommendation reliability. Pick one public name, apply it across Organization schema, titles, footer, llms.txt, LinkedIn, and directories (keep CVDH LLC as the legal entity).
Entity graph gap in sameAs: Organization schema currently lists only X/Twitter profiles (x.com/kaizenailab, x.com/dhoesq) as sameAs links — while the LinkedIn company page (linked visibly from /contact/) is not registered in the schema at all. Add LinkedIn, and any Crunchbase/directory profiles as they get created, to sameAs so search and AI engines can connect the entity across surfaces.
Policy decisions to confirm: (a) the GPTBot/ClaudeBot mismatch above; (b) blocking Google-Extended does not affect Google Search or the AI features governed by Googlebot, but it can limit certain Gemini-side uses. If maximum AI-surface presence is the goal, Don should confirm these trade-offs are intended.
No AI-visibility baseline exists. Live money-question probes (asking ChatGPT/Perplexity/Google AI Mode "who should a mid-market company hire for compliant AI implementation" etc.) could not be run this session — no browser surface available, and fabricating results is not an option. This needs a recurring monthly probe set; see action plan.
Proof-point density: copy leans on positioning claims; AI recommendations favor verifiable specifics (client counts, deployment numbers, timelines, named outcomes). Add quantified case studies.
Third-party validation: the visible external footprint is thin — one LinkedIn company link (on /contact/) and two X profiles in schema. Directory presence, industry coverage, and citations are what AI engines cross-check before recommending a firm. This is the slowest lever, so it should start early.

Prioritized action plan

#	Action	Impact	Effort	Time to results	Owner
1	Fix `/book` CTA on 3 industry pages (repoint to /contact/ or Cal.com); add 301 for /book	High	Low	Immediate	Mason
2	301 the 4 legacy pillar URLs → /guides/*; update the 9 in-content links pointing to them (across /author/don-ho/ and 3 guide pages), plus the separate readiness-framework link — confirm its intended target	High	Low	1–2 weeks	Mason (target confirm: Jen)
3	Link /services/ from primary nav + guides/industry pages (de-orphan)	High	Low	2–4 weeks	Mason
4	Remove /hero-preview/ from the XML sitemap	Med	Low	Immediate	Mason
5	Standardize brand entity: "Kaizen AI Lab" vs "Kaizen AI Labs" everywhere; add LinkedIn (currently missing) alongside the existing X profiles in Organization `sameAs`	High (GEO)	Medium	30–60 days	Don (decision) / Mason (implementation)
6	Rewrite titles + meta descriptions on commercial pages (home, about, services, guides, industries, web-labs), then top articles by impressions	Med	Medium	2–6 weeks	Mason (copy: Jen review)
7	Add ContactPage schema to /contact/ and CollectionPage to /industries/	Med	Low	2–4 weeks	Mason
8	Homepage H1 decision: single H1, demote second brand block to H2	Low	Low	4–8 weeks	Jen/Don (design call)
9	Restore Ahrefs API access (currently 403 — likely plan/entitlement) so backlink, keyword-gap, and Brand Radar modules can run	High (unblocks measurement)	Low	Immediate	Don / Carson
10	Stand up the monthly GEO probe: fixed money-question set across ChatGPT, Perplexity, Google AI Mode; log mentions/citations vs competitors	High (GEO)	Medium	60–90 days to trend	Jen + Mason
11	Add 2–3 quantified case studies / proof-point blocks to services + industry pages	Med (GEO)	Medium	60–90 days	Jen (content) / Mason (build)
12	Confirm AI-crawler policy: GPTBot and ClaudeBot (both training crawlers) currently allowed despite training-blocked policy; Google-Extended / FacebookBot blocking intended?	Med	Low	n/a	Don

Scope, method, and limitations

In scope this run: full crawl of all 101 sitemap URLs (titles, metas, headings, canonicals, robots, schema, internal/external links, word counts, response times), redirect and TLS verification, broken-link sweep with source attribution, sitemap and robots.txt analysis, llms.txt review, schema inventory, GEO readiness assessment, external CTA endpoint checks.
Not measurable this run (disclosed, not estimated): Core Web Vitals lab/field data (PageSpeed API rate-limited, HTTP 429 — a tooling limitation, not a site failure); backlink profile, domain rating, keyword rankings (Ahrefs API returned 403 Forbidden — access/entitlement issue to resolve, action #9); Search Console impressions/CTR (no property connection in this environment); live AI-answer money-question probes across ChatGPT, Perplexity, and Google AI Mode (no browser surface available this session — this is the AI-visibility baseline module and the single most important fast follow for the GEO half of this audit). None of these were fabricated; all are marked as open items. This report is therefore a full observable-site SEO/GEO readiness audit with disclosed measurement gaps, not a complete visibility/performance audit.
Method note: the crawl reads static HTML (server-rendered output), not a JS-executing browser. The site is an Astro static build, so rendered-vs-source divergence risk is low, but any purely JavaScript-injected element would not be visible to this crawl (relevant to the /services/ orphan caveat above).
Module fit: the local-SEO modules of the 21-module system (Google Business Profile, map pack, NAP citations) were assessed as largely not applicable — kaizenailab.com is a national B2B consulting property. If Kaizen maintains a GBP listing, say the word and a GBP teardown gets added.
Prior-audit comparison: no prior artifact found in any connected system; this report is the baseline of record, archived for future diffs.