Technical field guide AI Search Readiness

What business owners should know

Your website may rank in Google and still be invisible to AI.

AI search visibility is not one trick, one plugin, or one score. Before ChatGPT, Perplexity, Gemini, Claude, Google AI, or Copilot can recommend a business, the site has to be crawlable, readable, extractable, trusted, and measured engine by engine.

Built from public crawler documentation, search fundamentals, and Alchemetics field testing.

Audit question: “Can answer engines read, trust, and cite this business?”
1
Crawl access
robots
2
Raw HTML
readable
3
Answer extraction
weak
4
Trusted sources
gap
OpenAIGooglePerplexityClaudeBing

The mental shift

SEO gets you found. AI search decides whether you are worth recommending.

Traditional SEO was built around ranking position and clicks. AI search adds another layer: citation and recommendation. A page can rank, get crawled, become a candidate source, feed part of an answer, and still never be named. That is why Alchemetics audits the whole path, not just the ranking.

1

Crawl access

Can the right bots reach the pages, or are robots rules, redirects, security tools, or bad status codes blocking them?

2

Readable HTML

Does the important content exist in the initial HTML response, or is it hidden behind JavaScript that some crawlers may not render?

3

Extractable answers

Does each page answer real buyer questions clearly enough that a passage can stand alone inside an AI answer?

4

Entity trust

Does the web clearly agree who the business is, what it does, where it operates, and why it should be trusted?

5

Citation proof

Are there credible third-party sources, reviews, directories, mentions, and proof pages that support the recommendation?

6

Engine measurement

Are ChatGPT, Gemini, Perplexity, Claude, Google AI, and Copilot tracked separately instead of blended into one vague score?

Step 1

Make sure AI crawlers can access the site.

The first failure is often silent. A robots.txt file may allow Google but block an AI retrieval bot. A firewall may challenge a crawler even when robots.txt says it is allowed. A redirect chain may work for a browser but fail for a faster crawler. The audit starts by checking access bot by bot.

Check
What can go wrong
What Alchemetics looks for
Robots.txt
Important AI search bots are blocked or treated the same as training crawlers
Clear access rules for the bots that can support visibility
CDN and firewall
Cloudflare or security rules challenge crawlers before they see the page
Status checks from crawler-style requests, not just browser checks
Redirects
Long chains waste crawl budget or fail before the final page
Clean one-hop redirects wherever possible
Bing discovery
Important pages are missing from Bing, which can affect AI discovery
Bing Webmaster Tools, sitemaps, and IndexNow where appropriate

Step 2

Make the page readable without relying on JavaScript.

Many modern sites look complete to humans but expose very little useful text in the raw HTML. That creates risk for AI retrieval systems, partner search indexes, and tools that do not fully render JavaScript. The safest approach is server-rendered or statically rendered content where the important words are present immediately.

The simple test

Fetch the page as raw HTML. Turn JavaScript off. Compare that text against the rendered browser page. If the headline, service copy, FAQs, proof, and internal links disappear, the site is not AI-search ready.

Fix the source, not just the design

Use server-side rendering, static generation, clean HTML, crawlable links, schema, and content that exists before scripts run.

Step 3

Write pages so answer engines can extract the right passage.

AI systems do not always use a whole page. They retrieve sections, compare sources, and assemble answers. That means each important section should work by itself: direct answer first, clear entity names, specific proof, useful structure, and no vague filler.

A

Answer-first opening

Start the page with a direct 40 to 60 word answer to the main buyer question.

Q

Question-shaped headings

Use headings that match how people ask AI tools for help, not only old keyword fragments.

C

Complete sections

Each section should make sense even if AI pulls it away from the rest of the page.

P

Proof density

Use dates, numbers, names, examples, reviews, case studies, and credible references.

S

Structured formats

Tables, lists, comparisons, FAQs, pros and cons, and best-for sections are easier to parse.

F

Freshness

Show current dates when pages are updated and refresh important answers regularly.

Step 4

Build trust beyond your own website.

Owned content matters, but AI engines often need corroboration. If competitors are supported by reviews, directories, Reddit threads, YouTube videos, trade publications, partner pages, and comparison lists, they may be easier to recommend. Alchemetics identifies those source gaps and turns them into a practical authority plan.

R

Reviews

Review volume, quality, freshness, and platform coverage can shape trust.

D

Directories

Trusted local, industry, and category directories can become source signals.

M

Mentions

Independent mentions help answer engines corroborate claims about the business.

V

Video

YouTube and other video sources can influence AI-shaped discovery in many markets.

P

Proof pages

Case studies, customer stories, comparison pages, and outcomes make claims easier to support.

C

Consistency

Names, addresses, service descriptions, and categories should match across the web.

Step 5

Measure each answer engine separately.

There is no single AI search engine. A business can appear in Gemini and be absent from Perplexity. It can be mentioned by ChatGPT for one prompt and ignored for another. The measurement layer should track the exact prompt, the exact answer, the competitors named, the citations used, and the movement over time.

Engine
What to watch
Practical lever
ChatGPT
Whether the business is named, summarized correctly, and supported by trusted sources
Bing visibility, authoritative pages, clear entity signals, and third-party proof
Perplexity
Citation sources, freshness, community and third-party references
Current content, trusted source gaps, reviews, Reddit or community presence when appropriate
Google AI and Gemini
Overlap with Google rankings, answer fit, topical authority, and credible sources
Classic SEO foundation plus answer-ready pages and schema
Claude
Whether structured pages and supplied sources explain the business clearly
Logical depth, clean formatting, direct answers, and well-organized proof

Public sources

Credibility starts with what the platforms actually publish.

AI SEO is moving quickly, so Alchemetics separates official platform documentation from vendor studies and marketing claims. The hard percentages in this market should be treated as directional until your own prompt tracking proves them in your category.

O

OpenAI crawler documentation

OpenAI publishes crawler information for GPTBot, OAI-SearchBot, and ChatGPT-User at developers.openai.com.

A

Anthropic crawler documentation

Anthropic documents ClaudeBot and related crawler behavior for site owners.

P

Perplexity crawler documentation

Perplexity publishes crawler and user-agent information for site owners.

B

Bing IndexNow

IndexNow gives site owners a way to notify participating search engines about changed URLs.

G

Google Search Central

Google’s SEO fundamentals still matter for crawlability, indexing, structured data, and page quality.

M

Market testing

The final ground truth is prompt tracking in the client’s actual category, not a generic industry score.

FAQ

Questions about AI search readiness.

What is AI Search Readiness?

It is the technical, content, entity, trust, and measurement foundation that helps answer engines read, understand, cite, and recommend a business.

Can a website rank in Google but fail in AI search?

Yes. Ranking is only one signal. A site can rank but still be hard for AI systems to extract, verify, or cite.

Should I block AI training crawlers?

That is a business decision. The key is not to accidentally block search or retrieval crawlers that may support visibility.

Do AI engines all use the same sources?

No. Different engines use different retrieval systems, indexes, source preferences, and citation behavior.

Is llms.txt required?

No. It is a low-cost forward-looking file, but it should not be sold as a current ranking factor or citation guarantee.

What is the first audit Alchemetics runs?

We start with access and readability: robots rules, status codes, redirects, raw HTML, schema, indexation, and whether the page answers real buyer questions.

Get the report

Find out whether AI can read, trust, and recommend your business.

The AI Visibility Guru report shows the real prompts, answers, competitors, sources, and first fixes that matter most.