Background
The Dress Outlet is a US-based women's fashion retailer with over 4,000 SKUs and a well-established organic search presence — ranking on page one for dozens of competitive fashion keywords. Despite strong SEO fundamentals, AI-referred traffic was near zero in January 2026.
We ran a full AEOvision AEO audit in late January. The initial score was 34/100. Here's what we found and what we fixed.
Issue 1 — GPTBot was blocked
The site's robots.txt included a blanket disallow for all bots not explicitly whitelisted. GPTBot was not on the whitelist. This meant OpenAI's crawler had never indexed the site's content for training or retrieval.
Fix: Added GPTBot, PerplexityBot, and Google-Extended to the robots.txt allowlist. Also created an llms.txt file at the domain root. This single change took 48 hours to take effect.
Issue 2 — Homepage BLUF score was 12/100
The homepage hero copy opened with brand history: "Welcome to The Dress Outlet, where we have been selling dresses since 1998..." This tells an LLM nothing useful about what the site sells, who it serves, or what makes it different.
Fix: Rewrote the hero copy to lead with a BLUF sentence: "The Dress Outlet sells formal and semi-formal dresses in sizes 0–30, priced from $49 to $299, with free shipping on orders over $75." The new BLUF score was 81/100.
Issue 3 — No FAQPage schema on category pages
Category pages like /evening-dresses had no structured data. These are exactly the pages LLMs look to when answering "where can I buy evening dresses" queries.
Fix: Added FAQPage JSON-LD to the top 20 category pages, each with 3–5 question-answer pairs covering size range, price range, shipping, and return policy.
Results at 30, 60, and 90 days
- Day 30: AEO score up from 34 to 61. First ChatGPT citations detected. AI-referred sessions: 2,400/month.
- Day 60: AEO score up to 74. Perplexity citations appearing. AI-referred sessions: 14,200/month.
- Day 90: AEO score up to 82. AI-referred sessions: 33,000/month. A 340% increase from baseline.
The three fixes — unblocking AI crawlers, rewriting BLUF content, and adding FAQPage schema — cost less than 20 hours of implementation time combined.