How Does Perplexity Shopping Pick Apparel, and How Do You Get Selected?

Short answer: Perplexity and Claude pick fashion brands from pages they can fetch and read at the moment of the question. Both run live web searches through their own crawlers, PerplexityBot for Perplexity and Claude-SearchBot and Claude-User for Claude, and both cite the pages they use. A brand gets picked when those crawlers can reach its pages and the pages state product facts plainly: materials, sizes, prices, care and who the product is for. A brand that blocks the crawlers, or hides facts in images and PDFs, does not get cited.

How Perplexity picks sources

Perplexity's crawler documentation says PerplexityBot "is designed to surface and link websites in search results on Perplexity. It is not used to crawl content for AI foundation models." A second agent, Perplexity-User, visits a page when a user's question needs it. Perplexity recommends allowing PerplexityBot in robots.txt to appear in results and notes that robots.txt changes can take up to 24 hours to apply.

For a fashion brand, that means Perplexity's shopping answers are built from what its crawler last read on your site and on the sites that write about you, such as retailers, reviews and trade press. A product page with fiber content, size range and price in plain text is easier to quote than a lookbook image with the same facts set in type.

How Claude picks sources

Anthropic's crawler page lists three agents. ClaudeBot collects training data. Claude-User fetches pages when a person asks Claude a question, and Anthropic says blocking it "may reduce your site's visibility for user-directed web search." Claude-SearchBot "navigates the web to improve search result quality for users," and blocking it prevents Anthropic from indexing your content for search.

A brand can block ClaudeBot to keep its pages out of training and still allow Claude-User and Claude-SearchBot so Claude can find and cite it. Each agent is controlled separately in robots.txt.

Citation rates by engine in our own tracking

The F* Word tracks 99 fashion workflow prompts, such as questions about tech packs, sampling and pre-production software, in four AI engines every day. Table 1 shows how often each engine cited a thefword.ai page between June 22 and November 3, 2026. These are our own results for one B2B fashion brand, not an industry benchmark, and they show how differently the engines behave on the same prompts.

The same pages earned a citation rate more than three times higher on ChatGPT than on Claude. A brand that measures only one engine can draw the wrong conclusion about its AI visibility. For ChatGPT-specific steps, read how to get a fashion brand recommended in ChatGPT.

What makes a fashion page easy to cite

The book The Machine-Readable Brand (Kumar, Sidhik and Lokam, 2026) names the shift: "Three kinds of machines now read a fashion company's product data: the factory systems that make the garment, the regulatory systems that certify it, and the shopping agents that increasingly sell it." A page that answers an AI engine well carries the same facts a factory needs, in text.

• Fiber content, weight and construction written out, not only in an image
• Size range, fit notes and measurements per size
• Price, availability and delivery regions
• Care instructions and country of origin
• A one-sentence statement of who the product is for

Those facts usually start in the tech pack and bill of materials (BOM). The F* Word generates a tech pack with BOM in 8 to 10 minutes from a garment design, so material, trim and measurement facts exist as one record before they are copied into product pages. For how agents rank products once they find them, read how AI shopping agents choose fashion products.

Check your site in five minutes

1. Open yoursite.com/robots.txt and confirm PerplexityBot, Claude-SearchBot and Claude-User are not disallowed.
2. Check your firewall or CDN bot rules, since a block there overrides robots.txt.
3. Pick three product pages and confirm fiber content, sizes and price appear as text.
4. Ask Perplexity and Claude the same buyer question about your category and note which pages each cites.
5. Repeat the same questions monthly, because answers change as engines recrawl.

Product data checklist for AI shopping answers

Perplexity and Google both read structured product data. Google Merchant Center documents the apparel fields it expects for clothing products.

The F* Word's own tracking of 99 prompts showed ChatGPT, Claude and Gemini citing thefword.ai far more often than Google AI Overviews. The material and size fields above come straight from the tech pack, which The F* Word generates in 8 to 10 minutes. See also how AI shopping agents choose fashion products.

Frequently Asked Questions

Does Perplexity use my site to train its models?

Perplexity's documentation says PerplexityBot is not used to crawl content for AI foundation models. It is used to surface and link websites in Perplexity search results.

Can I block Claude training but still appear in Claude answers?

Yes. Anthropic lists ClaudeBot for training, and Claude-User and Claude-SearchBot for user requests and search. Each can be allowed or blocked separately in robots.txt.

Is Claude less likely than ChatGPT to cite a brand?

In our own tracking, the same pages were cited in 40.1 percent of ChatGPT answers and 11.2 percent of Claude answers. The engines search and select sources differently, so measure each one separately.

How long until a robots.txt change shows up?

Perplexity says up to 24 hours for its systems to apply a change. Citations then depend on when the engine next fetches your pages.

Further Reading

• Merchandising and launch workflow
• What the agent shelf is
• How AI shopping agents choose fashion products

Start your product facts as one structured record: try The F* Word free.