GEO basics: get cited by AI answer engines

More shoppers now ask an AI assistant "which waterproof leather boot ships in Turkey" than type that into a search box. The assistant reads store pages as text and quotes the ones it can understand. Generative engine optimization, GEO, is the practice of being one of those stores. Kambloo treats it as a core feature, not an add-on.

Four things an AI engine needs#

1. Access. The crawler must be allowed in and must get full content without running scripts.
2. Structure. Headings, tables and lists it can parse.
3. Completeness. Enough facts on the page to answer a question.
4. Quotability. Short, self-contained blocks it can lift.

Kambloo handles the first two. The last two are mostly about how you write.

What Kambloo does automatically#

Server-rendered pages. Every page is complete in the first response. AI crawlers, which usually do not run JavaScript, see everything a human sees.

AI crawlers allowed. Well-known AI crawlers are allowed on live domains by default. If you have a legal or brand reason to block them, there is a switch in the SEO settings. On the test store they are blocked along with everyone else.

llms.txt. On your live domain, Kambloo publishes a summary file at /llms.txt, following the emerging convention for AI engines. It contains a short description of your store, the full category list with product counts, your top products by sales and reviews (100 by default), your policy pages with one-line summaries, and links to your product feed and sitemap. It is rebuilt with the sitemap. You do not edit it directly; you improve it by improving your pages.

Semantic HTML. Themes are required to use one main heading per page, proper heading order, landmark regions, real tables for specifications and real markup for prices. Themes that break these rules do not reach the Market.

Structured data. Products, offers, breadcrumbs, FAQs and your organization are described in a linked structured data block on every page.

Feeds. Kambloo publishes a Google Merchant feed, a Meta catalog feed and a plain JSON product feed, refreshed every six hours and on price or stock changes. Shopping assistants and some AI engines read merchant feeds directly. Connect the Google Merchant feed in Merchant Center for free product listings.

What you write#

Specification attributes. Fill in the category template attributes on every product you care about. Kambloo turns them into a visible table, structured data and feed fields from one source. "What is the sole made of?" gets answered from the table. Your admin shows a completeness score per product; use it as a to-do list.

FAQ blocks. Add the questions customers ask, and answer them in the first sentence. Kambloo suggests FAQs from your shipping and return policies; accept and edit them. FAQs are printed as visible content and as FAQ structured data.

First paragraphs. On products, categories and static pages, make the first paragraph a complete answer to "what is this and who is it for". AI engines quote first paragraphs.

Policy pages. Shipping, returns, payment and contact pages are what an assistant reads when asked "does this store deliver in two days" or "can I return it". Keep them factual and current. They also feed the policies section of llms.txt.

Consistent facts. If your announcement bar says free shipping above 1,500 TL and your shipping page says 1,000 TL, an assistant will pick one, and it may pick wrong. One fact, everywhere.

What you should not do#

• Do not paste hidden text for crawlers. Everything AI engines read is also what customers see.
• Do not write "AI-friendly" filler. Short factual answers beat long ones.
• Do not block AI crawlers "to be safe" unless you have a specific reason. A store that cannot be read cannot be recommended.

Measuring#

GEO measurement is early. Kambloo's health report includes a GEO group: low specification completeness, high-traffic products without FAQs, and a stale llms.txt. For now, judge by the health report and by whether your policy answers are correct.

Common mistakes#

• Specifications written into the description text instead of the attribute fields.
• FAQ answers that do not answer in the first sentence.
• Shipping and return pages that were never written, so llms.txt has empty policy lines.
• Different numbers for the same policy on different pages.

Apply this in your store: read your llms.txt, then fix the policy pages and complete attributes and FAQs on your top ten products.