How to Optimise a Webflow Site for AI Search (llms.txt, Schema, AEO)

What optimising for AI search actually means

People now ask ChatGPT, Claude, Perplexity or Google AI Mode a question and get one composed answer with a few links. If your page is one of those links, you get a visit. If your brand is named in the answer, you get awareness even without a click.

The names for this vary: AEO (answer engine optimisation), GEO, AI SEO. I use AEO. The useful thing to know is what the platforms themselves say. Google's documentation states that there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary. The same foundations apply: an indexable page, helpful content, clear structure.

So the honest order of work is: make the site crawlable, make the content easy to quote, then add the optional extras. Below I go through that in the order I would do it on a Webflow project.

Step 1: crawler access and robots.txt

Webflow generates a robots.txt for every site and lets you edit it in the SEO section of site settings. Editing needs a paid site plan. Note that robots.txt controls crawling, not indexing, and it is a request, not a lock. Webflow explains the basics in its robots.txt guide.

The important point is that AI companies run separate bots for separate jobs, so you can allow one and block another:

Sources: OpenAI bot documentation and Anthropic crawler documentation.

My default for a business that wants leads is to leave the search and user bots allowed. Whether to block the training bots is a business decision for the client. Write it down, because a blanket Disallow: / for all AI user agents is a common and expensive mistake. If a CDN or firewall sits in front of the site, check that it does not block the same bots without anyone noticing.

Step 2: llms.txt in Webflow

llms.txt is a plain-text file in your domain root that gives language models a curated overview of your site. Webflow added support in July 2025: you can upload an llms.txt file in the site SEO settings, and Webflow serves it from the root of your domain while keeping it out of search indexes.

1. Open site settings, then the SEO tab, and find the LLMs.txt section.
2. Upload a UTF-8 file. Webflow's help article describes a size limit of 100 KB and says it appears on the published domain, not on the staging domain.
3. Publish, then open yourdomain.com/llms.txt and check it.

What goes in it: one line on who you are, then links with one-line descriptions to your most important pages, such as services, pricing, key case studies and documentation. Keep it short and curated, not a dump of the sitemap.

Be realistic about the value. Webflow itself describes the file as experimental, and Google's guidance above does not mention it. I add it because it takes twenty minutes, but I would never promise a client results from it.

Step 3: structured data and schema

Google also says there is no special schema.org markup you must add for AI features. Schema is still worth doing for two reasons: it removes ambiguity about who you are and what a page is, and it supports normal rich results.

• Organization on the home page: name, logo, URL, social profiles.
• Article or BlogPosting on blog templates, with author and dates bound to CMS fields.
• BreadcrumbList where the page hierarchy is real.
• Product or Service only when the page really describes one.

In Webflow you add JSON-LD as custom code on a page or template, and CMS template pages can pull values from collection fields, so one template covers every item. Two rules: the markup must match what is visible on the page, and you test it with Google's Rich Results Test before publishing. Do not add FAQ markup to pages that have no real FAQ.

Step 4: structure content so it can be quoted

This is where you have the most control. Answer engines pull passages, not whole pages, so every section should survive being quoted on its own.

• Put the answer first. Under each heading, start with a one or two sentence direct answer, then explain.
• Write headings as the questions people ask. "How much does a Webflow redesign cost?" beats "Investment".
• Use one H1 and a clean H2/H3 hierarchy. Choose levels by structure, not by font size.
• Use lists and tables for comparisons. They are easy to extract and easy for readers to scan.
• Be specific. Names, numbers, dates and sources give a model something to cite. Vague marketing copy gives it nothing.
• Show who wrote it. An author, a date and an "updated" date on every article.
• Keep content in HTML text. Key facts hidden in images, sliders or content loaded only after interaction may never be read.

A TL;DR box at the top and an FAQ at the bottom are cheap ways to get this structure. Add the FAQ only with questions you can answer well.

Step 5: measure AI referrals

You cannot measure everything, but you can measure two things: visits that come from AI tools, and whether your brand appears in answers.

Referral traffic

In your analytics, look at referral sources for domains such as chatgpt.com, perplexity.ai, claude.ai and gemini.google.com. Create a saved segment or channel group for them, so the trend is visible month by month. Expect small numbers at first, and note that some AI visits arrive without a referrer.

If you use Webflow Analyze, Webflow's AEO analytics adds AI visitors shown separately by source, prompt tracking across ChatGPT, Claude, Gemini and Perplexity, and LLM bot insights from server logs. It requires a Team or Enterprise plan with the Analyze add-on, so it is not for every client.

Prompt testing

Without that tooling, do it by hand. Write 10 to 20 prompts your customers would really ask, run them monthly in the main assistants, and record whether you are named, cited or absent. Keep the list fixed, so the comparison is fair.

The checklist

1. Check robots.txt: search and user bots allowed, training bots a documented decision.
2. Check that no CDN or firewall blocks the same bots.
3. Confirm key pages are indexable and in the sitemap.
4. Upload a short, curated llms.txt and verify it on the live domain.
5. Add Organization schema, plus Article schema on blog templates, and test it.
6. Rewrite key pages so each section opens with a direct answer.
7. Add authors and dates, and refresh old articles with real updates.
8. Set up an AI referral segment in analytics.
9. Write down 10 to 20 test prompts and re-run them monthly.

My take: structure first, tricks later

I have been building in Webflow since 2018, and my honest view is that AI search rewards the same things good SEO always did: a clear structure, real content and a fast, crawlable site. I do not have my own controlled test data on llms.txt, so I will not claim it moves anything.

What I do know from client work is that structure is where Webflow is strong. On Big Moe Watches, a Dubai retailer built white-label via Landrieau Studio, we built a Webflow catalogue of 1,100+ luxury watches with advanced filtering, and every watch is a CMS item with consistent fields. That same discipline is what makes content easy for any machine, search engine or assistant, to read and quote. If you want the wider SEO groundwork, start with the Webflow SEO checklist, and for large collections see building a large CMS catalogue.

I also build AI features and automations for clients, so I am watching this area closely. If you want a technical AEO pass on a Webflow site, whether yours or a client's, get in touch.

FAQ

Written byPeter CekanakDesigner & Webflow developer · white-label partner for agencies

I design and build Webflow sites for agencies and startups since 2018 — Client-First, GSAP, CMS and AI automations. Sites I built helped a client to a $10M exit and tripled demo requests for a SaaS.

Webflow Certified ExpertRelume ExpertClient-First