Run all of this inside the full Agent OS — join the AI Profit Boardroom →

agentos.guide › blog

AEO For Ecommerce In Your Agent OS: Five Copy-Paste Prompts For Crawler Checks, Product Markup, Answer Briefs And Reviews

By Julian Goldie · 12 November 2026 · agentos.guide

Illustrated hero: a gold product card, a crawler path and an answer bubble linked on a dark aubergine background

AEO for ecommerce is mostly catalogue groundwork, and your Agent OS can do the repetitive half of it: fetching product pages, checking crawler rules, drafting markup and sorting customer questions.

Book a free SEO strategy session →

The rule for everything below is simple: the agent reads and drafts, and you publish.

Google's generative AI search guide treats AEO as ordinary SEO, so these prompts stick to documented fundamentals rather than tricks.

If you'd rather have my team check the agent's output with you, book a free SEO strategy session.

Why a store is a perfect job for an agent

A shop with 200 products has 200 pages built from the same template, which means the same checks repeated 200 times.

That is tedious for a person and trivial for an agent with a URL list.

What the agent can't do is know your products, so its drafts always come back to you for facts and sign-off.

Think of it as a tireless junior who checks every page the same way, never skips one, and never decides anything on its own.

The time you save goes into the part only you can do: telling it what is true about each product.

Quick start: set up the store workspace

  1. Create a folder called store-aeo with subfolders audit, markup, briefs and reviews.
  2. Export your product catalogue to products.csv: name, URL, SKU, GTIN, brand, price, currency, stock status.
  3. Export the last few months of customer emails or chat logs to questions.txt, with names and emails removed.
  4. Write one line at the top of your agent instructions: read-only on the live site, draft-only for anything published or sent.

Keep the catalogue export fresh: an agent working from last month's prices will draft last month's markup.

Now paste the prompts in order, and read each output file before moving to the next prompt.

What good output looks like

Before you run anything, decide what a finished file should contain, so you can spot a lazy answer fast.

If a file arrives without sources, send it back with one line: "Show the URL and the exact text for each finding."

Book a free SEO strategy session →

Agents get sharper when they know their work will be checked line by line.

Prompt 1: crawler and preview check

This catches the blocks that quietly keep a shop out of AI answers.

You are auditing [domain] for AI search access. Read-only: change nothing.
1. Fetch [domain]/robots.txt. For each of Googlebot, Bingbot, OAI-SearchBot,
   PerplexityBot and GPTBot, say whether product URLs are allowed or blocked.
2. For the first 20 URLs in products.csv, fetch each page and record:
   HTTP status, final URL, canonical tag, any robots meta tag or X-Robots-Tag
   header (flag noindex, nosnippet, max-snippet, NOARCHIVE, NOCACHE),
   and any element using data-nosnippet.
3. Note whether the product name, price and description are in the raw HTML
   or only appear after JavaScript runs.
Save audit/access.csv and list the five most serious problems in audit/access.md.

OpenAI's crawler overview says sites that opt out of OAI-SearchBot won't be shown in ChatGPT search answers, and that its settings for search and training are independent.

Perplexity's crawler page likewise recommends allowing PerplexityBot and its published IP ranges.

Google's robots meta tag spec confirms that nosnippet also keeps text out of AI Overviews and AI Mode.

Prompt 2: product page answer audit

Read questions.txt and group the questions by product.
For each of our 15 best sellers in products.csv:
1. List the five questions customers ask most, in their own words.
2. Fetch the product page and mark each question as answered clearly,
   answered vaguely, or not answered.
3. Note any fact on the page that conflicts with products.csv
   (price, stock, size, materials).
Write briefs/answer-gaps.md as a table. Do not write new copy yet.

Google describes AI Overviews and AI Mode running several related searches across subtopics to build one response, so unanswered buyer questions are the gaps that matter.

📞 Want a person to read your audit files?

Send us the agent's output and book a free SEO strategy session; we'll go through the crawler, markup and content gaps with you.

→ Book your free session

Prompt 3: draft Product markup from the catalogue

Using products.csv, draft JSON-LD for each product page following Google's
merchant listing documentation: Product with name, image, description, sku,
gtin, brand, and a nested Offer with url, price, priceCurrency and availability.
Rules:
- Use only values from products.csv or the visible page. Never invent ratings.
- Add aggregateRating only if the page visibly shows that rating and count.
- One Product per URL. Flag category pages instead of marking them up.
Save one file per product in markup/ and a checklist of fields still missing.

Book a free SEO strategy session →

Google's merchant listing guide lists name, image and an Offer with a price and currency as required, and recommends placing markup in the initial HTML.

Check each draft in the Rich Results Test before it goes near your theme.

If your store sells the same item in several colours or sizes, ask the agent to flag the variants separately, because Google supports dedicated variant markup for them.

Return policies belong at business level too: Google recommends nesting them under Organization markup rather than repeating them on every product.

Prompt 4: answer-first copy brief per product

For each product in briefs/answer-gaps.md, write a brief, not finished copy:
1. A two-sentence opening that says what the product is and who it suits.
2. Short answers to the five buyer questions, marked [NEEDS OWNER FACT]
   wherever a measurement, material or policy must be confirmed.
3. One suggestion for an original photo or short video.
Do not copy manufacturer descriptions. Save to briefs/[sku].md.

Google's guide asks for "non-commodity content", and the owner facts are what make a product page more than a copy of the supplier's text.

Prompt 5: review request plan, draft only

Draft a review request plan for our store. Draft only: send nothing.
1. Write one short post-delivery email asking every buyer for a review
   of the specific product they bought.
2. No discounts, gifts or free products in exchange for reviews.
3. Ask everyone, not only buyers you expect to be happy.
Save reviews/request-email.md for me to approve.

Google's review snippet guidelines ban fake reviews and require marked-up ratings to be visible on the page.

Prompt 6: feed and page consistency check

Once markup is live, the next failure point is drift between the page, the markup and your Merchant Center feed.

Read-only. For every URL in products.csv:
1. Fetch the page and read the visible price, currency and stock status.
2. Read the Product JSON-LD on the same page and record the same three values.
3. Compare both against the matching row in our feed export, feed.csv.
List every product where any of the three sources disagree in audit/drift.csv,
with the value from each source side by side.

Book a free SEO strategy session →

Google's product data page warns that combining website and feed data can cause inconsistencies through lag, especially when items sell out.

It suggests letting Merchant Center update its copy from your site when it notices a mismatch, which is worth switching on before you rely on this prompt alone.

Run it after every sale or price change, because those are the days drift appears.

A weekly workflow that keeps it running

  1. Monday: rerun Prompt 1 on any URLs changed last week.
  2. Tuesday: run Prompt 4 for three products and fill in the owner facts.
  3. Thursday: publish the approved pages and markup yourself.
  4. Friday: check Search Console's generative AI performance reports and Bing's AI Performance report, then log the numbers.

A garden tools shop doing three products a week has its 40 best sellers rewritten inside a quarter.

Keep a simple log of what changed on which date, so a jump or dip in the reports can be matched to a specific batch of pages.

Without that log, you will be guessing which change made the difference.

What never to hand the agent

And skip llms.txt as a Google tactic, because Google says Search doesn't use those files.

When to hand it to a pro

The prompts cover groundwork, but a large catalogue with variants, multiple currencies and a Merchant Center feed needs someone who owns the whole plan.

No one, agent or agency, can guarantee a product shows up in an AI answer, as Google's hiring guidance makes clear.

Goldie Agency works on SEO, GEO, AEO and link building, Julian's own published numbers list 1,000+ SEO and link building campaigns, and the first step is a free strategy session.

Book a free SEO strategy session and bring your audit files, so the call starts with your data.

For the platform-specific version, see generative engine optimisation for Shopify.

🔥 Learn AI SEO yourself and build these agents properly

Inside the AI Profit Boardroom you get step-by-step video trainings, the installable Agent OS these prompts run in, and four live coaching calls a week.

→ Join the AI Profit Boardroom

📘 Two free resources

Grab the 200+ AI SEO Prompts for more audit and brief prompts, and the free link building book for earning links to product guides.

→ Or learn AI SEO yourself in the Boardroom

Book a free SEO strategy session →

Also On Our Network

FAQ

Can my Agent OS do AEO for my shop on its own?

It can do the groundwork: crawler checks, page audits, markup drafts and copy briefs. You still supply product facts and publish every change yourself.

Which prompt should I run first?

Prompt 1, the crawler and preview check. A blocked crawler or a stray nosnippet tag cancels out every other improvement.

Will the agent make up ratings in the markup?

Not if you tell it not to. Prompt 3 only allows values from your catalogue file or the visible page, and you check every draft in the Rich Results Test.

Should the agent send review requests automatically?

No. Keep it draft-only, approve the email yourself and never offer gifts or discounts in exchange for reviews.

Final word on AEO For Ecommerce In Your Agent OS

Run Prompt 1 today, fix what it flags, and when the catalogue outgrows your week, book a free SEO strategy session and let my team take over your AEO for ecommerce.

About Julian

I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.

I help business owners scale with AI agents, automation and SEO.

My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.

→ Get my best AI training inside the AI Profit Boardroom

Related guides on agentos.guide

→ Generative Engine Optimization For Shopify In Your Agent OS

→ AI Overview Optimization Services: Prep Work In Your Agent OS

→ Brand Visibility In AI Search: Track It In Your Agent OS

→ AI Visibility Audit: 5 Agent OS Prompts To Run It

Get the whole Agent OS

The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.

Join the AI Profit Boardroom →

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉

← all guides on agentos.guide