AI / GEO Readiness

Verified: Katy vs. The Woodlands scored 97.3% textual similarity; the flagged Woodlands duplicate pair scored 98.5%. With this much overlap, AI crawlers and embedding models will likely treat these as near-duplicate documents — diluting which single URL (if any) an AI system would choose to cite for "tree service in [city]," and wasting crawl attention that could go toward a smaller number of genuinely distinct, citation-worthy pages.

Status (Oct 4, 2026, updated three times): real progress, not a full fix. Round 1: a genuinely unique ~80-100 word paragraph was added under a new question-style H2 on all 43 real city pages (99.8% → 91.0%). Round 2: varied the "Why Choose Us" 3-blurb section's wording (same real facts, previously identical) and added a second new paragraph using real Houston-region facts (clay soil, climate, hurricane season) (91.0% → 76.0%). Round 3 (client kept pushing): varied the 4 remaining service-category blurb cards (previously 100% identical across all 43 pages), same real claims, different wording. Re-measured live: Klein vs. Katy similarity now 59.5% (was 76.0%, originally 99.8%). What remains is mostly shared nav/footer/CTA chrome (normal for any site) plus the bio/credential sidebar box, which needs real local specifics (landmarks, job photos, testimonials) from the client to go further. Sulphur LA and Lake Charles LA were removed entirely at the client's request; a third out-of-territory page, Georgetown TX (~170 miles away, missed in the original audit), was found and noindexed.

Extracted every paragraph from the homepage, a city page, a service page, and the tips page. City-page paragraphs run 4–42 words (median ~29); zero fall in the 134–167-word band AI systems favor for direct extraction. The site's body copy is short marketing-voice sentences, not self-contained factual answers.

Status (Oct 4, 2026): same root cause as GEO-03, now improved alongside it. One real, factual answer-first paragraph was added to the homepage, /tree-removal-services/ turned out to already have solid content, and all 43 city pages now have a new ~80-100 word unique paragraph each (see GEO-03) — closer to the citable band than before, still short of the 134–167-word target. The rest of each city page's body copy is still short marketing-voice boilerplate.

/llms.txt and /llms-full.txt both 301-redirect to the homepage — Rank Math's generic catch-all, not a real file. Separately, /tree-maintenance-tips/ has genuine FAQ content ("Why Should You Hire An Arborist?", "What Is A Certified Arborist?") but zero FAQPage schema anywhere on the site — a zero-cost missed opportunity, since the content already exists.

Status (Oct 4, 2026): FAQPage schema added to /tree-maintenance-tips/ using the page's own real Q&A text, no new content invented — verified live. llms.txt correction: the master AI-SEO playbook this agency follows explicitly says not to prioritize llms.txt — no major AI platform currently uses it. Leaving the Rank Math redirect as-is is the correct call, not a gap.

Question-based headings (which AI systems strongly favor for direct-answer extraction) appear in exactly one place sitewide: the "Frequently Asked Questions" section on the tips page. Everywhere else: "TREE REMOVAL", "SERVING CYPRESS TEXAS, HOUSTON AND SURROUNDING AREAS". Naturally list-friendly content (a PROS/CONS breakdown on the removal page) is written as plain paragraphs instead of scannable list items — a missed structuring opportunity.

Status (Oct 4, 2026): a real question-style H2 ("Looking for a Tree Service in [City]?") with a direct answer-first paragraph now exists on all 43 city pages, plus one on the homepage — 44 new question-style headings sitewide, up from 1. The original ALL-CAPS marketing headers (hero titles, "SERVING... AND SURROUNDING AREAS") are untouched, and a full rewrite of those is a real content project, not something to rush — especially since several obvious candidates (cost, licensing/insurance) have no real underlying facts on file to answer honestly yet.