Why AI recommends some D2C brands and ignores others: a five-part AEO strategy

Ask ChatGPT for "a linen shirt for a beach wedding in Goa, under ₹3,000" and it names two or three brands. Ask again next week and the names are often the same. The brands that keep getting picked are not lucky. AI assistants can read their store, find them on the pages they trust, and match them to the exact question the shopper asked.

This guide turns that into a plan. It has five parts, in the order a D2C team should do them, and a 30-day schedule at the end.

The short version

• Understand query fan-out. One shopper prompt becomes many searches. You need to cover the whole buying decision, not one keyword.

• Find your gaps. List the prompts where AI names other brands, or names you without linking to you.

• Research prompts. Build them from seed words and modifiers, then sort them by whether AI answers them in full.

• Earn citations. Fix your own pages, get named on other sites, publish video, and keep AI crawlers allowed.

• Measure. Track AI referrals, AI bot visits and what buyers tell you at checkout. Re-run the same prompts every week.

1. Understand how one prompt becomes many searches

A search engine matches one query to a list of pages. An AI assistant does more work. Google says AI Overviews and AI Mode may use a "query fan-out" technique: they issue multiple related searches across subtopics and data sources, then build one answer from what comes back.

For the linen shirt prompt, the assistant might look up linen against cotton in humid weather, the dress code for a beach wedding, linen shirt brands in India under ₹3,000, how linen shirts fit, whether linen wrinkles, and delivery and returns to Goa. It then reads the best sources for each and writes a short answer with two to five products.

This changes what "ranking" means. A product page that only targets "linen shirt" answers one of those searches. The brand that gets recommended has answers for most of them: fabric facts on the product page, a fit guide, a clear returns policy, and a guide that compares fabrics for hot weather. Cover the decision, not the keyword.

The foundation is still ordinary search. Google says there are no additional requirements to appear in AI Overviews or AI Mode beyond normal Search eligibility. Pages that are crawlable, fast and clearly written are the starting point for every engine.

2. Find your gaps before you write anything

Most teams start by publishing. Start by measuring instead. Run 25 to 30 shopper prompts for your category in ChatGPT, Gemini and Google AI Overviews. Use logged-out or temporary sessions so past chats do not change the answer. For each prompt, record the brands named and the pages cited.

You will find three kinds of gap:

• Citation gap. The assistant names your brand but cites someone else's page, such as a marketplace listing or a review site. The shopper's click goes elsewhere.

• Competitor gap. A competitor is named for a prompt where you are not. These are the prompts that cost you sales today.

• Topic gap. The assistant links a topic you should own, such as "hypoallergenic earrings" for a jewellery brand, to another brand, or to no brand at all.

Write every gap down with the prompt, the engine and the date. You cannot fix them all at once. The list decides what you do in parts three and four. Our step-by-step guide to a free AI visibility audit shows how to run the prompts and what to record.

3. Research prompts, not only keywords

Shoppers type short keywords into Google. They ask AI assistants full questions with a budget, an occasion and a constraint. Keyword research still works, but you have to turn keywords into prompts.

Start with seed words and modifiers. Seeds are one or two words that describe what you sell, such as "linen shirt", "kurta set" or "vitamin C serum". Modifiers turn a seed into a real question: "best", "for", "vs", "under ₹2,000", "for sensitive skin", "for a wedding", "alternative to". You can ask an AI assistant to draft them with a prompt like this:

"I run a D2C brand that sells [products] to [audience] in [country]. Give me 10 seed terms of one or two words, and 10 modifiers that turn them into shopping questions about budget, occasion, comparison and problems. The seeds and modifiers should not share words."

Combine seeds and modifiers into prompts, then sort each prompt with one question: can the AI answer this fully without a click? "How do I wash a linen shirt" is answered in full, so the goal there is to be the brand named in the answer. "Best linen shirts under ₹3,000" ends in a shortlist, so the goal is to be on the list, with a link to your product page. Prompts with "best", "vs", "alternative" and "for [use case]" are where shortlists get made. Start there.

4. Earn citations in four places

Your own pages

AI engines read your pages in pieces. A section may be pulled out on its own, so each one should make sense without the rest of the page. Four habits help:

• Answer first. Put the answer in the first sentence of each section, then the detail. "Our linen shirts are 100% European flax, 160 GSM, and ship in 2 to 4 days" beats three lines of brand story.

• Name things. Write the product name, material, size range, price and use case. "Relaxed-fit white linen shirt, 100% linen, S to 3XL, ₹2,490" gives an engine facts it can repeat. "Our best-seller is perfect for summer" gives it nothing.

• One idea per sentence. Short, plain sentences are easier to quote correctly. If a sentence needs two reads, split it.

• Keep it current. Update prices, stock, sizes and policies when they change, and refresh guides that have gone stale. A page with a wrong price is a page an engine learns not to trust.

Comparison pages, buying guides and "best for" lists are the formats that end in a shortlist. If shoppers in your category ask "X vs Y", write the honest comparison yourself. Our guide on how to get your products recommended by ChatGPT covers product facts, Product schema and feeds in detail.

Mentions on other sites

AI answers lean on what other people say about you. When the same facts about your brand appear on many trusted pages, an engine is more confident repeating them. Three kinds of page matter:

• Editorial pages: gift guides, "best of" lists, reviews and trade press in your category. They are the hardest to earn and the most often cited.

• Communities: Reddit, Quora and category forums where shoppers ask for advice. Answer real questions with real detail. Do not post ads.

• Your own channels: Instagram, LinkedIn, YouTube and marketplace listings. Make sure the facts match your site: the same prices, sizes, materials and policies everywhere.

Look at the cited pages you logged in part two. They are your outreach list.

Video

Video is a format AI answers can cite, and it shows things text cannot: how a fabric drapes, how a serum absorbs, how a ring sits on the hand. Make videos that answer search questions rather than chase trends: a fit guide, a fabric comparison, a "how to choose" explainer. Put the question in the title and the first line of the description. Add chapters. Say the product names out loud, because spoken words become part of the transcript.

Crawl access

None of this matters if AI crawlers cannot read your store. Most AI companies run separate crawlers for search answers, for model training and for fetching a page when a user asks. Blocking the search crawler removes you from that engine's answers.

Open yourstore.com/robots.txt and look for Disallow rules under OAI-SearchBot, Claude-SearchBot, PerplexityBot or Googlebot. Stores can block AI crawlers by accident, through an old template, a plugin or a CDN setting. Cloudflare began blocking known AI crawlers by default on new domains in July 2025, so check your CDN settings as well as the file. Cloudflare's managed robots.txt also adds a Content-Signal line, and its default says AI training is not allowed. If you want to be used for training as well as cited, set that yourself.

A simple file that welcomes AI engines looks like this. It is the one eldor.ai uses:

User-agent: *
Content-Signal: search=yes, ai-input=yes, ai-train=yes
Allow: /
Sitemap: https://eldor.ai/sitemap.xml

5. Measure what you can, and ask about the rest

AI search does not have a Search Console with impressions and positions. Three signals together give a clear enough picture:

• AI referral traffic. ChatGPT adds utm_source=chatgpt.com to the links it shows in search answers. In GA4, add a custom channel group for referrers such as chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. Treat the number as a floor: copied links and some in-app clicks arrive as direct traffic.

• AI bot visits. Your server or CDN logs show which pages OAI-SearchBot, Claude-SearchBot and PerplexityBot fetch. Pages that search bots fetch again and again are likely being used as sources.

• What buyers tell you. Add "How did you hear about us?" to checkout or a post-purchase survey, with "AI assistant (ChatGPT, Gemini and others)" as an option. Many shoppers see your name in an AI answer, then search for you on Google. Analytics calls that organic search. Only the buyer can tell you otherwise.

Then re-run your prompt list every week and track your answer share: how often you are named, how prominently, and how accurately. Weekly is the right rhythm. AI answers vary from run to run, and one day tells you little. Our answer-share scorecard shows how to report it like a performance channel.

Your 30-day plan

• Week 1: Check robots.txt and your CDN for AI crawler blocks. Set up the AI referral channel in GA4. Add the "How did you hear about us?" question.

• Week 2: Run 25 to 30 shopper prompts in ChatGPT, Gemini and Google AI Overviews. Log every citation, competitor and topic gap.

• Week 3: Fix the biggest gap on your own pages. If AI misstates your products, fix product facts and schema. If a competitor wins comparison prompts, write the comparison page. If old guides lost their rankings, refresh them first. That is often the fastest win.

• Week 4: Pitch one editorial list or gift guide, answer five community questions, and publish one search-led video. Re-run the prompts and compare with week 2.

After that, keep the weekly prompt run and pick the next biggest gap each month. AEO sits on top of SEO. It does not replace it, so keep the technical basics healthy as you go. If you are still deciding how the two relate, see AEO for D2C brands: how answer engines find, choose and quote your products.

Where Eldor fits

Eldor does parts two, four and five for D2C stores. The free Eldor AI audit runs 25 shopper prompts on ChatGPT and Google AI Overviews, checks 100 of your products against your live store, and gives you a fix plan ranked by revenue. On paid plans, Eldor tracks your prompts every week by market (US, Europe and India) and by plan (logged out, free and paid). It also checks that what AI says about your products matches your store, and fixes product data and policies so engines can quote them. See agent readiness for the details.

FAQ

What is an AEO strategy?

An AEO strategy is a plan to get your brand named and linked in AI answers from ChatGPT, Gemini, Perplexity and Google AI Overviews. It covers how AI engines search, where your brand is missing, which prompts to target, how to earn citations, and how to measure progress.

How long does AEO take to work?

Fixes to your own pages can show up once engines crawl them again. Mentions on other sites and video take longer to build. Re-run the same prompts every week and judge the trend over one to three months, not a single day.

Should I block GPTBot?

GPTBot is OpenAI's training crawler. Blocking it does not remove you from ChatGPT search answers, which use OAI-SearchBot. Block OAI-SearchBot and ChatGPT cannot cite your pages in search. Most D2C brands that want AI to recommend them allow both.

Is AEO different from SEO?

AEO builds on SEO. Google says AI Overviews and AI Mode have no extra requirements beyond normal Search eligibility, so crawlable, well-structured pages come first. AEO adds work SEO does not measure: covering the whole buying decision, earning mentions on the pages AI cites, and tracking whether AI names you correctly.

How do I know if ChatGPT sends me traffic?

Look for utm_source=chatgpt.com and the chatgpt.com referrer in your analytics. ChatGPT adds the tag to links in its search answers. Count it as a floor, because copied links and some app clicks show up as direct traffic. Add a "How did you hear about us?" question to catch the rest.

Does longer content get cited more by AI?

Length on its own is not the goal. Engines pull short, self-contained sections that answer one sub-question clearly. A focused page with current facts, specific product details and an answer at the top of each section is more useful to an engine than a long page that buries the answer.

Keep reading: AEO for D2C brands · How to get your products recommended by ChatGPT · The answer-share scorecard · Free AI visibility audit

Sources: Google Search Central, "AI features and your website"; Google, AI Mode update at Google I/O 2025; OpenAI, "Overview of OpenAI crawlers" and the Publishers and developers FAQ; Anthropic Help Center on Anthropic's crawlers; Cloudflare, "Giving users choice with Cloudflare's new Content Signals Policy" and its July 2025 announcement on blocking AI crawlers by default. The approach is inspired by independent analysis of how AI engines choose sources.