The host-level AI-crawler blocklist has been partially lifted. ClaudeBot, PerplexityBot and Perplexity-User all return HTTP 200 today, where last week ClaudeBot had its connection killed outright and both Perplexity agents got a 403. That restores Claude and Perplexity as citation sources and is worth four points on Crawlability. GPTBot and CCBot are still killed at the edge, so OpenAI's index crawler and Common Crawl remain locked out.
The second change is the byline. Ten of the twelve most recently modified posts now carry "author":{"name":"Dinos Rologas"} in the Article schema, with zero instances of the old "Warner Pool Care Team" string. That was the single named E-E-A-T defect in last week's report and it is now substantially fixed.
Uncapped score this week is 103. The six categories can arithmetically total 120, so the displayed figure is capped at 100.
Dinos Rologas as the Article author node. This was the root-cause defect recorded in the project knowledge base — Yoast reading the WordPress post-author account rather than the intended person node — and the WordPress-side author has evidently been reassigned.post-sitemap.xml, most recent published 10 August. Question-format titles throughout — "Is Winter the Best Time to Buy a Spa in Brisbane?", "How Long Should I Run My Pool Pump in Winter?", "How Often Should I Change My Spa Water?" — each targeting a real Brisbane seasonal query."telephone": "07 3882 1722" in schema, a matching href="tel:0738821722" link, and all four branch addresses rendered in the page body — Albany Creek, Brendale, Warner and the Spa Supa Centre.Not a new finding, and not a regression. This blocklist was documented before the 10 August run. What changed is that it has been narrowed, not removed. robots.txt is still a Yoast allow-all block (User-agent: * / Disallow:) that the server disagrees with.
| Crawler | Result on / | Change vs 10 Aug |
|---|---|---|
| GPTBot/1.1 | connection killed | unchanged — still blocked |
| ClaudeBot/1.0 | HTTP 200 | fixed — was killed |
| PerplexityBot/1.0 | HTTP 200 | fixed — was 403 |
| Perplexity-User/1.0 | HTTP 200 | fixed — was 403 |
| CCBot/2.0 | connection killed | unchanged — still blocked |
| anthropic-ai | connection killed | unchanged (legacy agent, low impact) |
| Google-Extended | connection killed | unchanged (training opt-out only, no search impact) |
| Bytespider | connection killed | unchanged (ByteDance, low impact) |
| OAI-SearchBot/1.0 | HTTP 200 | unchanged — allowed |
| ChatGPT-User/1.0 | HTTP 200 | unchanged — allowed |
| Googlebot/2.1 | HTTP 200 | unchanged — allowed |
The practical cost is now narrower than it was. ChatGPT can still reach the site live through OAI-SearchBot and ChatGPT-User, so ChatGPT answers about Warner Pool Care are not blind. What GPTBot's block costs is inclusion in OpenAI's own pre-built index, and CCBot's block costs inclusion in Common Crawl, which feeds a long tail of downstream models.
Requests must be spaced 12–15 seconds apart or results are unreliable. Every result above was captured with 14-second spacing.
/robotic-pool-cleaner-winter-leafy-brisbane-block/ still returns "author":{"name":"Warner Pool Care Team"} with four instances of the string in the page. /blog-posts/ returns an empty author name. These are stragglers from an otherwise completed fix, not a systemic failure.
| Page | Blocks | Types found |
|---|---|---|
| / | 5 | Organization, WebSite, WebPage, BreadcrumbList, ImageObject, LocalBusiness ×4, Service ×3, Person ×2, FAQPage |
| /is-winter-best-time-to-buy-a-spa-brisbane/ | 5 | Article, Organization, WebSite, WebPage, BreadcrumbList, ImageObject, LocalBusiness, Service, Person, FAQPage |
No malformed JSON-LD detected on any page audited.
/zzz-fake-12345 returns HTTP 404 with a distinct document (md5 b736bf84 vs homepage 5b4ae7a8). No catch-all routing.sitemap_index.xml present with 7 child sitemaps: post (99), page (30), footer_builder_pt, category, post_tag, author, ecstore.meta name="author" tag, and in the visible byline.| # | Action | Impact | Effort |
|---|---|---|---|
| 1 | Ask the host to remove GPTBot and CCBot from the edge blocklist, matching the ClaudeBot/Perplexity change already made | High | Low |
| 2 | Fix the author on /robotic-pool-cleaner-winter-leafy-brisbane-block/ — reassign the WordPress post author to Dinos | Medium | Low |
| 3 | Fix the empty author node on /blog-posts/ | Medium | Low |
| 4 | Add AggregateRating schema to back the "100+ 5-Star Reviews" claim | Medium | Low |
| 5 | Split the paragraph-length homepage H2s into a short heading plus body copy | Medium | Low |
| 6 | Add internal links from each blog post into the matching service or store page | Medium | Medium |
| 7 | Bring robots.txt into line with what the server actually enforces | Low | Low |
| 8 | Expand service page copy toward the 600+ word mark to match blog depth | Medium | Medium |
| 9 | Add dateModified to Article schema on older posts | Low | Low |
| 10 | Add a Spa Supa Centre-specific FAQ block targeting spa purchase queries | Medium | Medium |
Warner Pool Care hits the display cap this week on an uncapped 103. Two things moved: the edge blocklist was narrowed so Claude and Perplexity can now crawl, and the blog byline was reassigned to Dinos Rologas across almost the whole library. Both were named as the specific blockers in the 10 August report, and both have been acted on.
What remains is a short list. GPTBot and CCBot are still blocked at the host, which is the same one-line fix that evidently already worked for ClaudeBot. Two posts kept the old byline. Neither is urgent, and neither is holding the displayed score down — the site has three points of headroom above the cap.