On this page
The patterns in every audit#
We’ve run Visibly audits across a wide range of sites — B2B SaaS, DTC e-commerce, marketplaces, content businesses. The mistakes below appear with surprising consistency. None are exotic. None require expensive tooling. Most take less time to fix than to read.
1. Burying the answer below 200 words of marketing copy#
The page’s URL is /what-is-x, the implied query is “what is X?”, and the first three paragraphs are positioning copy. Move the actual answer to the first sentence under the h1. Easy fix: rewrite the lead to open with a clean, declarative definition.
2. Missing datePublished and dateModified#
Without explicit dates in Article schema, freshness-sensitive surfaces (Perplexity, AI Overviews) treat the page as undated and silently down-weight it. Easy fix: add both to your Article JSON-LD; bonus points for displaying the updated date visibly in the article header.
3. Stale “Updated 2023” dates on otherwise good pages#
When you do a real content refresh, update dateModified in the JSON-LD and the visible date in the page header. Easy fix: automate it so a content update always bumps both.
4. Blocking AI crawlers in robots.txt by accident#
Some CMSes inherit blanket User-agent: * disallow rules from default templates. Audit your robots.txt explicitly for OAI-SearchBot, PerplexityBot, and Bingbot. Easy fix: add explicit Allow: / rules under each.
5. Publishing to a subfolder content mill#
Hosting AI-generated content on yoursite.com/articles/ or content.yoursite.com/ is transparent to both Google and AI systems. Easy fix: publish into the main site structure, not an isolated content folder. Visibly’s CMS adapters do this by default.
6. No FAQPage schema on pages with obvious Q&A blocks#
If your page already has questions and answers, marking them up with FAQPage schema is a 10-minute win. Easy fix: the Schema.org FAQPage reference gives you the exact structure; copy-paste and adapt.
7. Skipped heading levels#
Going from h1 to h3 without an h2 confuses extraction. Easy fix: enforce one h1, h2s for major sections, h3s only as children of h2s.
8. No alt text on images#
Models do extract image alt text and use it for context-disambiguation. Easy fix: run a 15-minute alt-text pass across your top 10 pages.
9. Generic anchor text on internal links#
Linking to a page with anchor text “click here” or “learn more” leaves citation signal on the table. Easy fix: use the page’s actual title as the anchor when it makes sense in the prose.
10. No internal linking between adjacent topics#
Models triangulate authority by looking at how densely your site links between related pages. Easy fix: once a quarter, review every cornerstone page and add 2–3 contextual internal links to related pieces.
11. Publishing without a sitemap#
If you have a blog or a content section, you need a sitemap and you need to submit it to Bing Webmaster Tools (for ChatGPT’s fallback discovery) and Google Search Console. Easy fix: most CMSes auto-generate a sitemap; just submit it.
12. Treating thin posts and cornerstones the same#
Forty thin 600-word posts dilute your cluster authority more than they help. Easy fix: delete or consolidate the thin posts; double down on 6–8 cornerstones that actually answer commercial queries.
What to do after the easy fixes#
The hard work (items 5, 9, 10, 12) is structural — it takes weeks, not hours. That’s where the audit-driven content engine pays off, because the audit tells you specifically which cornerstones to build and which thin posts to consolidate.
Run the free audit and you’ll get the prioritized fix path specific to your site. Or subscribe to Visibly Monitor at $99/mo to start tracking AI visibility across all five surfaces while you do the work.
Related questions worth a look
- How do I check if AI crawlers can access my site?
- Why does ChatGPT cite some sites and not mine?
- Is AI-generated content bad for SEO?
- What schema markup does ChatGPT actually use?