dardo
ESLet’s talk ↗

AI search

How to prepare a website for AI search

Appearing in AI answers starts with a clear offer, accessible pages and verifiable evidence. Treat access, content and measurement as separate tasks; no file or tag guarantees a recommendation.

Download the observation log ↓
01

Make the useful answer available in HTML

A search assistant needs access to the page that answers a question. Publish the service, audience, deliverables and supporting project in readable HTML. Keep the title, heading, visible copy and structured data consistent. A visual animation can introduce a brand, while a separate service page gives the offer a stable URL.

02

Check access before adding another file

Check robots.txt, noindex directives, authentication and edge challenges on the important URLs. OpenAI distinguishes its search crawler from its training crawler; Perplexity also documents a search crawler. An HTTP request using a bot name is a useful smoke test, but does not prove access from the provider’s actual network. Inspect verified bot traffic when it becomes available.

03

Write comparisons that someone can verify

Name the providers, explain the selection and link the source beside the claim. Separate a published service description from your interpretation of project fit. If your studio publishes the list and appears in it, say so before the reader evaluates the options. A table and concise questions help people compare the same facts.

04

Measure citations, visits and inquiries separately

Bing’s AI Performance report shows sampled citation activity in supported Microsoft AI experiences. Referral analytics measure people who click through. Inquiry attribution connects some of those visits to a received project brief. None is a complete count of every AI recommendation. Keep the provider, date, prompt, cited URLs and outcome in separate fields.

05

What llms.txt can and cannot establish

An llms.txt file is an optional guide to a site’s important resources. Dardo generates it from the same public content used by the website so the service and project facts stay aligned. The proposal is not a promise that an assistant will read it, cite it or rank the business. Keep the normal website complete for people and search engines.

06

A practical baseline before the next change

Use a fixed set of real buying questions in English and Spanish. Record the answer date, provider, sources and whether Dardo is mentioned or cited. Leave an unrun question blank. Repeat observations before attributing a change to your work, and compare against qualified inquiries rather than treating a single answer as a stable ranking.

Download the observation log ↓

Sources & further reading

Sources behind this guide, with further detail from the original publishers.

How we publish these guides →
From an idea to something real

Make your next
move matter.

Start a project ↗Explore the project planner →