A shop owner asked me last month: “Can you rank my store at number one in ChatGPT?” Then she added, “and Claude and Perplexity too.” I understood the hope. Customers now ask AI tools things like “best running shoes for flat feet under ₹5,000”, and the answer shows a few products and links.
My honest reply was: nobody can sell you a guaranteed spot, because these systems do not publish ranking factors. What you can do is remove the reasons you would be skipped and add the reasons you would be chosen. That is what AI SEO for e-commerce really is.
How these tools find products (the part you can verify)
Claude. Anthropic’s help centre describes three crawlers. ClaudeBot collects content that may contribute to model training. Claude-User visits a page when a person’s question needs it. Claude-SearchBot looks at content to improve the relevance of search responses. Anthropic says its bots respect robots.txt, and that blocking Claude-User or Claude-SearchBot may reduce your visibility in Claude’s search answers.
Perplexity. Perplexity describes PerplexityBot as the crawler that surfaces and links sites in its search results, and says it is not used for training foundation models. It recommends allowing it. Perplexity-User fetches pages when a user’s action needs them and generally ignores robots.txt, because the user triggered the request.
DeepSeek. Here I have to be careful. I could not find official DeepSeek documentation about how its web search retrieves sources or which crawler it uses. Third-party trackers list a “DeepSeekBot” as reported, not operator-verified. So I would not build a DeepSeek-specific strategy on guesses. The advice below, which is good for Google and for every other AI tool, is the sensible approach for DeepSeek too.
Step 1: Do not block the doors by accident
The cheapest win is also the most common failure (it is pure technical SEO, and it applies to AI tools as much as Google). Check three places:
- robots.txt: are you disallowing Claude-SearchBot, Claude-User or PerplexityBot? Some security or “block all AI” plugins do this by default.
- Your firewall or CDN (WAF): bot protection can silently block legitimate crawlers. Perplexity advises allow-listing by matching the user agent and its published IP list together. Anthropic publishes IP ranges too.
- Rendering: if prices, specs and reviews appear only after heavy JavaScript runs, a fetcher may see an empty page. Make sure the core product facts are in the HTML.
Whether you allow training crawlers such as ClaudeBot is a business decision. Many stores allow the search and user-fetch bots and make a separate choice about training. Decide it deliberately, not by default.
Step 2: Make product facts unmistakable
AI answers lean on clear, consistent facts. For every product page I would check (our product page optimisation guide goes deeper):
- A unique title and a plain-language description: who it is for, what it does, what it is made of, sizes and variants.
- Specifications in a visible table, not buried in images.
- Price, currency and availability, kept accurate and matching your structured data.
- Product schema with name, image, brand, SKU or GTIN, offers, and review ratings where genuine.
- Shipping, returns, warranty and payment information in text, on the page or one click away.
- Your product feed in Google Merchant Center kept clean, because accurate feeds help in Google and give you a tidy source of truth for everything else.
Perplexity has been building shopping features, and third-party guides describe a merchant programme. I could not check its current terms on Perplexity’s own site, so please read Perplexity’s official merchant information before acting on any eligibility or cost claims you see elsewhere.
Step 3: Write the content AI tools actually quote
People ask comparison and decision questions, not just product names. Build pages that answer them honestly (more on this in how to get cited by ChatGPT and Perplexity):
- Buying guides: “how to choose a [product] for [situation]”, with real criteria.
- Comparison pages: your product versus common alternatives, including where you are not the best fit. Honest comparisons are cited more naturally than sales copy.
- Use-case pages: “best [product] for beginners / small kitchens / monsoon”.
- FAQ blocks on product and category pages answering real customer questions: sizing, compatibility, care, delivery times.
Use short, direct answers near the top of each section, then detail. Write with first-hand detail: your own tests, measurements, photos, customer feedback. [ADD: one real ClicZeo or client e-commerce example, such as a page where clearer specs or a buying guide improved enquiries or impressions.]
Step 4: Build trust outside your own site
AI systems cross-check. A brand that appears consistently in reviews, forums, marketplaces, news mentions and trusted directories looks safer to cite than one that exists only on its own website. So:
- Encourage genuine customer reviews on the platforms your buyers use. Never fake them or pay for positive ones.
- Take part in relevant communities honestly, answering questions without spamming links.
- Keep brand name, address, contact details and policies consistent everywhere.
- Earn coverage through useful data, guides or original photography.
Step 5: Keep it fresh and maintain it
Out-of-stock products with live “buy” buttons, old prices and discontinued models erode trust, with people and with machines. Update stock, prices and dates, and show “Updated on” for guides. Remove or redirect dead pages cleanly.
What I would not bother with yet
- Chasing special files. Files like llms.txt are discussed a lot, but I have not seen official confirmation from these three platforms that they rely on them. Add one if it costs little, but do not expect it to move anything. We looked at the evidence in does llms.txt help AI Overviews?.
- Prompt-stuffing. Hiding instructions for AI in your pages is unreliable and risks being treated as manipulation.
- Mass-produced AI pages. Thin, scaled content is a bad bet for Google and for AI citations alike.
How to measure something you cannot rank
Because there is no rank tracker you can fully trust, measure in layers:
- Server logs: do the Claude and Perplexity user agents visit your key pages, and do they get a 200 response?
- Referral traffic: in analytics, watch visits and orders from perplexity.ai and claude.ai sources.
- Prompt testing: keep a list of 20 to 30 real customer questions, ask them monthly across the tools, and note whether your brand or pages appear. Record the date, because answers change.
- Search Console: Google’s AI features report impressions too, which tells you whether your content is useful there.
The bottom line
You cannot buy a number-one slot in Claude, DeepSeek or Perplexity. You can make your store reachable, factual, well-reviewed and easy to quote. Do that, and you are better placed in every search surface, old and new. If you are weighing paid options too, compare SEO vs Google Ads ROI and the new ChatGPT ads pricing.
If you would like a practical audit of your store’s AI visibility through our e-commerce SEO or GEO services, talk to the ClicZeo team.
