Continuation of the Backyard Paradiso GEO + SEO audit. Part 1 covers the executive summary, score breakdown, and critical/high/medium-priority issues. ← Read Part 1 first.
06AI citability — 48/100
Citability scores how likely AI systems are to cite passages from a page. Research from Princeton and IIT Delhi (2024) found AI systems preferentially cite passages that are 134–167 words, self-contained, fact-rich, and answer a question in the first 1–2 sentences. (Directional research — treat as hypothesis.)
| Category | Weight | Score | Comment |
|---|---|---|---|
| Answer Block Quality | 30% | 42 | Sections open with marketing prose, not direct answers. |
| Self-Containment | 25% | 58 | Most blocks name their subject; some over-rely on “we”/”our”. |
| Structural Readability | 20% | 68 | H1/H2/H3 hierarchy intact, FAQ section, lists used. No tables — would help. |
| Statistical Density | 15% | 22 | Almost no numbers. No costs, no install timelines, no warranty years, no wind-load ratings. |
| Uniqueness | 10% | 35 | Generic copy that could apply to any aluminum pergola installer. |
Strongest block on the Pergolas page
The “Why Choose an Aluminum Pergola?” section with its four sub-features (Durability / Low Maintenance / Eco-Friendly / Aesthetic Appeal) is the closest the page gets to citable content. Each feature has a concrete claim attached. It’s still missing numbers — “resists rust, corrosion, and warping” is a claim AI systems will accept but not preferentially cite over a competitor saying “tested to ASTM B117 salt-spray standard for 1,000 hours.”
Weakest block on the Pergolas page
The rewrite earns: definition pattern in sentence one, specific facts (seven markets, 50+ year rating, 2–4 day install), self-containment, and 60-word standalone extractability.
Per-page citability snapshot
| URL | Citability | Biggest gap |
|---|---|---|
| Pergolas | 52 | No numbers, marketing-prose openings |
| Artificial Grass | 38 | Fence content collision; weak openings |
| Decks | 34 | Fence H2 embedded; Chinese alt text; no numbers |
| About Us | 31 | Aspirational prose, no facts about the company |
| Homepage | 18 | Almost no body content |
| Blog post (Pergolas vs Gazebos) | 62 | Best on the site — has comparison table, pros/cons, definitions. |
07Brand authority — 18/100
Brand authority is the weakest category in the audit. The Ahrefs December 2025 study suggests brand mentions on AI-cited platforms correlate ~3x more strongly with AI visibility than backlinks.
| Platform | Weight | Score | What we found |
|---|---|---|---|
| YouTube | 25% | 28 | Channel exists at @BackyardParadiso, but no third-party video coverage or installer-channel mentions. |
| 25% | 5 | No mentions found in any DIY, landscaping, or homeowner subreddit search. | |
| Wikipedia / Wikidata | 20% | 0 | No Wikipedia article. No Wikidata entity. |
| 15% | 8 | No company page found in search. | |
| Other (Yelp, BBB, Angi, press) | 15% | 22 | Facebook and Instagram active. No surfaceable Yelp/BBB/Angi pages. |
What to build first (90-day brand-authority sprint)
- Google Business Profiles for all seven locations. Target 30+ reviews per location in 90 days through systematic post-install review requests.
- LinkedIn company page with all seven locations as office records, employees linked, weekly post cadence focused on completed projects.
- YouTube content cadence — 1–2 install timelapses per month per market, properly titled with location-tagged metadata.
- Local press in 2–3 markets — pitch local home & garden sections.
- Authentic Reddit participation — no brand seeding. Months of work, not weeks.
08Content E-E-A-T — 38/100
December 2025 Quality Rater Guidelines updates extend E-E-A-T to all competitive queries, not just YMYL topics. AI systems are increasingly distinguishing between content that reports on a topic and content from someone who has done it.
| Dimension | Weight | Score | Comment |
|---|---|---|---|
| Experience | 25 | 8/25 | No project case studies, no first-person install accounts, stock photos throughout. |
| Expertise | 25 | 9/25 | No author credentials, no technical depth, generic terminology. |
| Authoritativeness | 25 | 5/25 | No external citations, no media coverage, no awards, no industry partnerships listed. |
| Trustworthiness | 25 | 16/25 | NAP visible across seven locations, HTTPS valid. Duplicated testimonials hurt the score. |
Highest-leverage content moves
- Real project case studies — one per service per market over 12 months.
- Author profile + Person schema for whoever writes the blog.
- Trust strip in footer or About — manufacturer partnerships, insurance/license info, certifications.
- Add specifics to every service page — warranty years, wind-load ratings, install timelines, material certifications.
09Technical foundations — 55/100
Some categories couldn’t be verified directly because the audit environment can’t fetch robots.txt, sitemap.xml, or measure Core Web Vitals against the live origin.
| Category | Score | Notes |
|---|---|---|
| Crawlability | 9/15 | Sitemap presumed via WordPress default. AI crawler access needs explicit verification. |
| Indexability | 5/12 | Cannibalization on category archives confirmed. Index bloat likely. |
| Security | 7/10 | HTTPS valid. HSTS, CSP, X-Frame-Options not verified. |
| URL Structure | 5/8 | Service URLs clean. Product subpages at root path inconsistent. |
| Mobile | 7/10 | Viewport meta tag present. Tap targets and font sizes need real-device audit. |
| Core Web Vitals | 7/15 | Cannot measure without CrUX or Lighthouse access. |
| Server-Side Rendering | 12/15 | WordPress is server-rendered. Schema not in raw HTML. |
| Page Speed | 3/15 | Many large images, Chinese-character filenames suggest no optimization workflow. |
Recommended verification steps before signing off on technical
- Pull live
robots.txtand confirm GPTBot, ClaudeBot, PerplexityBot, Bingbot, Googlebot, Google-Extended, CCBot, Amazonbot all explicitly allowed. - Pull
sitemap.xmland verify URL count, lastmod dates, that all service and blog URLs are listed. - Run PageSpeed Insights on the homepage, Pergolas page, and one blog post.
- Check IndexNow integration (instant Bing notification).
- Check Google Search Console for indexing status, mobile usability, schema errors.
10Schema & structured data — 8/100
This is where the highest-leverage technical improvement lives. As of the audit, no JSON-LD blocks are visible to crawlers fetching raw HTML on any of the eight pages sampled.
Schema priorities for this business
- Organization schema — homepage. Required for entity recognition. Include name, url, logo, description, foundingDate, sameAs, contactPoint, knowsAbout, areaServed.
- LocalBusiness schema × 7 — one block per location, deployed on the location pillar pages once built.
- Service schema × 14 — one per service page.
- FAQPage schema — every service page that has an FAQ section.
- Article + Person schema — every blog post.
- BreadcrumbList schema — matching the visible breadcrumbs.
- Product schema — the 11 artificial-grass product variants and any other product-style sub-pages.
Implementation path: install RankMath Pro, use the Schema Generator to attach types globally and per-page, validate every output through validator.schema.org and Google’s Rich Results Test before declaring a page done.
11Platform readiness — 42/100
Platform optimization scores how ready each page is for the major AI surfaces. Different platforms favor different signals; only ~11% of domains are cited by both ChatGPT and Google AI Overviews for the same query.
| Platform | Score | Biggest gap |
|---|---|---|
| Google AI Overviews | 38 | Pages don’t rank top-10 for service-city queries. AIO sources 92% from top-10 organic. |
| ChatGPT (Bing index) | 32 | No Wikipedia entity, no Reddit signals, weak entity recognition. |
| Perplexity | 45 | Pages have decent structure but lack the statistical density and named sources Perplexity weights. |
| Claude | 52 | Cleanest structure of the four. Improves further with proper schema and author identity. |
| Bing Copilot | 42 | Same as ChatGPT (shared Bing index). IndexNow not detected. |
12Quick wins — this week
Seven moves that take under five hours of total work
- Fix the fence-content leakage on /services/artificial-grass/ and /services/decks/. 30 minutes.
- Replace the testimonials block on the homepage, About Us, Contact, and Artificial Grass pages. 1 hour.
- Bulk-fix Chinese filename alt text on the Decks page and the rest of the site. 1 hour.
- Install RankMath Pro and deploy Organization schema on the homepage. 30 minutes.
- Add a 60-word company description to the homepage above-fold. 30 minutes total.
- 301 the WordPress category archive URLs to the corresponding service pages. 30 minutes.
- Add a static “Updated [Month YYYY]” line to all 14 service pages. 30 minutes.
1330-day roadmap
Week 1
All seven Quick Wins above. In parallel: pull live robots.txt and sitemap.xml, run PageSpeed Insights on all sampled URLs, claim or audit Google Business Profiles for all seven locations.
Week 2
Schema rollout phase 1: Organization (homepage) + Service schema for all 14 service pages + FAQPage schema on every page with an FAQ section. Begin bulk image alt-text remediation.
Week 3
Build the seven location pillar pages at /locations/[city]/. Each gets H1, address, map embed, services list with internal links, location-specific testimonials, LocalBusiness schema, service-area copy.
Week 4
Author identity rollout — assign a real human to the blog, build the author page with Person schema, retroactively re-byline existing posts. Generate and publish llms.txt. Begin GBP review-collection sprint.
Beyond 30 days the work shifts to brand-authority and content-volume territory: 10–15 cluster posts per service silo, one project case study per service per market over the year, YouTube content cadence, LinkedIn company page launch, local press outreach in two priority markets, schema completeness for all blog posts and product pages, ongoing review-volume growth on GBP.
14Methodology & caveats
Audit methodology follows the geo-seo skill (v2.0), composite-scoring across six categories: Citability (25%), Brand Authority (20%), Content E-E-A-T (20%), Technical (15%), Schema (10%), Platform (10%).
Pages sampled: homepage, /about-us/, /contact-us/, /blog/, /services/pergolas/, /services/artificial-grass/, /services/decks/, /pergolas-gazebos-pavilions-whats-the-difference/, /pergolas/.
Limitations of this audit:
robots.txt,sitemap.xml, andllms.txtcouldn’t be fetched directly. Verify in step one of remediation.- Core Web Vitals weren’t measurable without Lighthouse / PageSpeed Insights access.
- Schema was tested by inspecting raw HTML returned by web fetcher. JS-injected schema wouldn’t show in this view.
- Brand authority scoring used external search reconnaissance only.
Methodological caveats from the geo-seo skill itself:
- The 134–167 word optimal-citation length and the +40% / +115% citation-lift figures come from individual academic studies. Directional, not load-bearing physics.
- The Ahrefs Dec 2025 “brand mentions outweigh backlinks 3x” finding is one study on 75K brands. Platform weights are derived from that single study.
- Wikipedia presence is recommended for ChatGPT and Perplexity citation, but the editorial notability bar makes it unrealistic for an SMB outdoor-living installer in the short term.
- Reddit recommendations should be authentic participation only. Brand-account posting is detected and removed.