Fixed-price audit · ₹12,500 +91 89254 44011 WhatsApp 6 am – 11 pm
Free tool · no email gate

The AEO Readiness Checklist — Score Your Site Out of 100 for AI Search Visibility

Twenty checks, five points each, five sections. It takes about twenty minutes with your site open in one tab and Search Console in another, and it will tell you which of the five things is actually costing you AI citations.

In short

Score your site across 5 sections worth 20 points each: answer-first structure, schema coverage, entity consistency, directory presence and crawler access. Schema alone is a 2.3× citation lever and HowTo markup 2.8×. Under 40 means invisible to AI search; 71–100 means you are a default source in your niche.

Reviewed 20 September 2026 · re-scored against the published citation research each quarter; point weightings change only when the underlying studies do

The AEO readiness score: 5 sections × 20 points 100 points total Answer-first structure Does each page answer its question in 40–60 words? Schema coverage The 2.3× citation lever; HowTo is worth 2.8× Entity & NAP consistency One description, repeated verbatim everywhere Directory & platform presence GBP is 67% of AI Overview local citations Crawler access & llms.txt GPTBot, ClaudeBot, PerplexityBot, Google-Extended Score 0–40: invisible. 41–70: occasionally cited. 71–100: a default source for your niche.
Five sections, twenty points each. The weighting is even on purpose: in practice a site with 20/20 on schema and 0/20 on crawler access scores zero citations, because the door is shut.

How to use this

Open your site and score each of the twenty checks below out of five. Five means done properly across the whole site; three means done on some pages; zero means not done. Add the five section totals. The whole thing takes about twenty minutes and needs nothing you have to pay for — Search Console, a schema validator and your own robots.txt.

There is no email gate and no PDF to download in exchange for your details. If you want the scoring sheet as a spreadsheet, take the tables below; they are the tool. We would rather this page got cited than that it collected addresses, and given that pages citing named sources inline are cited 2.1× more often, the whole thing is built to be quoted.

Section 1 — Answer-first structure (20 points)

The block an engine can lift without rewriting. This is the cheapest section to fix and the one most Chennai sites fail outright.

CheckPointsWhat good looks like
Every important page answers its own question in 40–60 words, near the top5Not a welcome paragraph. A direct answer containing at least two real numbers, in the words the question is asked in.
Headings are phrased as the questions buyers actually type5“How much does SEO cost in Chennai per month” beats “Our pricing philosophy” every time.
At least one table or structured list of real figures per page5Tables are extracted readily. A page of adjectives is not.
Visible last-updated date and a stated review cadence5Recency showed no measurable correlation in the AI Overview study — but buyers check, and a 2023 stamp on a price list reads as abandonment.
Pages over 2,500 words are cited 1.6× more than sub-800-word pages, with the lift starting around 1,800 words.

Section 2 — Schema coverage (20 points)

The single biggest engineerable lever: 2.3× more citations in an analysis of 1,000 AI Overviews, rising to 2.8× where HowTo markup fits the content.

CheckPointsWhat good looks like
Organization and LocalBusiness (or ProfessionalService) sitewide, with one @id5One entity node referenced everywhere, not a fresh copy per page.
FAQPage on every service and locality page5Questions that match the page's own FAQs, not boilerplate pasted across the site.
Service, Product or Offer with a real price where you sell something5priceRange or an actual price. “Contact us” is not a price and cannot be marked up as one.
HowTo on anything procedural, BreadcrumbList throughout5HowTo is worth 2.8× on procedural content — checklists, processes, setup guides.
Validate every type. Invalid markup is worth nothing and a wrong price in schema is worse than none.

Section 3 — Entity and NAP consistency (20 points)

BrightLocal identified 415 distinct directories feeding AI answers. If they disagree about who you are, the model hedges — and a hedge is a competitor's name.

CheckPointsWhat good looks like
One canonical description of the business, repeated verbatim everywhere5Same 40 words on the site, GBP, Justdial, Sulekha, LinkedIn, IndiaMART. Word for word, not paraphrased.
Name, address and phone identical across every profile you hold5Including the door number format and the phone's spacing. Models match strings, not intentions.
Named people with real bios and consistent titles5A founder or lead with the same name and role on the site and on LinkedIn. This matters double in YMYL sectors.
Dead and duplicate profiles claimed or removed5An abandoned 2019 listing with an old number will be retrieved as readily as your current one.
This section is unglamorous and it is where most of the quick wins are.

Section 4 — Directory and platform presence (20 points)

Business websites earn 59.9% of local AI citations, but the identity layer decides whether you are eligible at all.

CheckPointsWhat good looks like
Google Business Profile complete: categories, services, photos, posts5GBP is 28.63% of all citations and 67% inside Google AI Overviews. It is not a side task.
Review velocity, with replies5Steady recent reviews beat a big old total. Reply to all of them, briefly.
Listed on the directories your buyers' answers draw from5For Chennai local that is Justdial and Sulekha; for agencies GoodFirms, Clutch and Semrush Agency Partners; for SaaS, G2.
A named human present on the platforms AI reads5Reddit, LinkedIn and YouTube are among the most-cited domains across AI search. Answer real questions; do not spam.
ChatGPT takes 60.8% of its local citations from Yelp — worth knowing even where that is not your market.

Section 5 — Crawler access and llms.txt (20 points)

The section that is worth zero points to get right and minus everything to get wrong.

CheckPointsWhat good looks like
GPTBot, ClaudeBot, PerplexityBot and Google-Extended allowed in robots.txt5Check each token by name. Defaults on some platforms block all four.
Key content renders without JavaScript5If the answer only exists after hydration, assume it does not exist.
llms.txt published at the root and kept current5Cheap insurance. An hour to write, no engine guarantees to read it, and it costs nothing to have.
No CDN, firewall or bot rule silently blocking AI user agents5Test from outside. A robots.txt that allows and an edge rule that blocks is the commonest false pass.
We find at least one blocked AI crawler on roughly half the sites we audit, including sites paying another agency ₹40,000 a month.

Interpreting your score

0 – 40 · Invisible

AI systems cannot reliably retrieve you or resolve who you are. Fix in this order: crawler access, then the entity layer, then one answer-first page. The first two are days of work, not months, and they unlock everything else.

41 – 70 · Occasionally cited

You appear on low-competition prompts and vanish on anything contested. The gap is almost always schema depth and informational assets — commercial pages average only 3.1 citations against 5.6 for definitional ones, so build the definitional ones.

71 – 100 · A default source

You are in the pool for your niche and the work shifts to holding position: fresh dates, new questions answered before competitors think of them, and a monthly citation-rate measurement so you notice a slide before your enquiries do.

How to run your own 10-prompt citation test

Scoring the checklist tells you what is wrong. A citation test tells you whether it matters. Write down ten questions a buyer would type into a chatbot before buying from you — not keywords, questions. Run each one ten times, in a fresh logged-out session each time, on ChatGPT and Perplexity at minimum. Record how many of the ten runs name you. That fraction is your baseline.

Do it logged out, with location set to where your buyers are, and do not read your own answers charitably: a mention of your sector without your name is a zero. If you want the comparison, our own grid of 30 prompts across 5 engines is published with the losses included, so you can see what a two-month-old domain scores against the same standard.

Checklist questions

How long does the checklist take?

About 20 minutes for a small site with Search Console open, and half a day for a large one where the schema and profile layers have to be checked properly. The 20 checks are 5 points each across 5 sections, so a section takes roughly 4 minutes if you already know your own setup.

Is schema really worth that much?

An analysis of 1,000 AI Overviews found pages with schema cited 2.3× more often, the largest engineerable effect in the study, and HowTo markup on procedural content worth 2.8×. Domain authority correlated more strongly overall at +0.61, but you cannot ship domain authority this quarter and you can ship schema this week.

What is a good score to aim for?

Seventy-one out of 100 is the threshold where sites tend to become a default source for their niche. Most Chennai business sites we audit score between 25 and 45, and the missing points are concentrated in schema coverage and crawler access rather than in content quality.

Do I need all five sections, or can I pick one?

Section 5 is not optional — a blocked crawler makes the other 80 points worthless. After that, the highest-return order for a site starting near zero is crawler access, entity consistency, schema, answer-first structure, then platform presence, because the first three are quick and the last two are ongoing.

Does word count matter for AI citations?

Somewhat. Pages over 2,500 words were cited 1.6× more than sub-800-word pages, with the lift beginning around 1,800 words and plateauing near 3,500. That is a reason to answer a question completely, not a reason to pad — page speed, recency and reading level showed no measurable correlation at all.

Can you run the checklist for me?

The AI-crawler and schema sections are inside the ₹12,500 technical audit, which covers 42 named checks and is delivered in 10 working days. The full entity, directory and platform work is part of the ₹60,000 SEO + AEO tier. The checklist itself is free and always will be.

Why is there no email gate on this?

Because a gated PDF cannot be cited, and pages that cite named sources inline are cited 2.1× more often than those that do not. This page is built to be quoted by an AI engine and by anyone writing about AEO. Collecting 200 email addresses would be worth less to us than one citation that keeps arriving.

How often should I re-score?

Quarterly, and immediately after any site migration or platform change. Migrations are where crawler access and schema silently break — we have seen all four AI user agents blocked by a default rule the morning after a replatform, with nobody noticing for six weeks.

Next steps

Send the URL. Get a tier, a price and a start date.

One working day, on WhatsApp, from a person who has already opened your site in Search Console-shaped eyes. No discovery call required before you get a number.