Have AI answers recommend your company
An AEO grader scores how ready your pages look to answer engines such as ChatGPT, Perplexity and Gemini. We check those pages, then ask your buyers' questions without your name and read whether the answers recommend you. A person on our team checks each page change before buyers see it.
What is an AEO grader?
An AEO grader examines a page or site for features an answer engine can use, then assigns a score. Can a crawler reach the page? Is the answer easy to find near the top? Other checks cover structured data, llms.txt and markdown copies. Together, the findings describe what an assistant can read on the site and identify parts of the existing pages that may need work.
The findings give us page work to do before adding more content. If a crawler cannot reach a buyer guide, writing another article leaves that access problem unresolved. If the useful answer sits far down a page, we can bring it forward. We handle those fixes with the buyer questions in view, so the pages explain what those buyers came to find.
The page gets a grade, but a buyer gets an answer to a particular question. To connect the two, we ask questions that leave your company name out and read the replies. The comparison is whether assistants name you when a buyer asks about the problem you solve. That makes the question itself part of the test, alongside the pages an answer cites.
- Score the page. A grader checks the material an engine can read and gives you findings about its readiness.
- Ask about the company. A question naming your business shows how an assistant describes a company the buyer already knows.
- Test discovery through questions. We ask buyer questions without your company name on three engines and record whether the answers introduce your company.
A grade alone costs you buyers asking AI
A good score can feel like the work is finished. Buyers may still receive answers that send them to another company.
- Lose the recommendation. An assistant can read your page and choose a rival whose case it finds more persuasive.
- Leave rivals familiar. Companies already established in an assistant's answers can become harder to displace as buyers keep encountering them.
- Miss outside influence. What other sites say about your market can shape answers beyond anything measured on your own pages.
Asking the buyer's questions shows where to investigate those missing recommendations.
We grade your pages and test the answers
We run the checks, read the replies and make the page changes. You can see which finding led us to each piece of work.
Check page access
We inspect up to 100 public pages for technical problems, including crawl access, schema, llms.txt and markdown copies, and retain the findings.
Write buyer questions
We turn the problems buyers research into questions that leave your company name out, so the test starts where discovery starts.
Read each answer
We record mentions, citations and recommendations across ChatGPT, Perplexity and Gemini, including the competing companies each answer puts in front of buyers.
Improve cited material
We work on the pages the answers rely on, adding the company's own proof where buyers need a reason to consider it.
We change pages only after reading the answers
The replies guide the fixes: we first see where assistants pass over your company, then examine the relevant pages.
First
Choose the questions
We write about 30 questions from your buyers' perspective, leaving your name out so each asks about a need in your market.
Then
Record the recommendations
We put the questions to three engines and note where your company appears, which rivals appear and which pages the answers cite.
Next
Work on missing pages
We check the pages that should help answer those questions and make the changes, with a person reviewing each one.
Each week
Repeat the answer test
After changes, we ask the same questions again and discuss which page work to continue, change or stop on the call.
Example: asking 30 buyer questions about one market
In October 2026, we asked ChatGPT, Perplexity and Gemini 30 buyer questions for a software startup. Its name appeared in 1 of 90 answers; its main competitor appeared in 43. Reading the answers showed which company a buyer would hear about. That gave us a different finding from a score of the startup's pages.
A fair test leaves your company's name out
A buyer asking how to solve a problem may never have heard of your company. We write the test from that position, so it shows whether an assistant introduces you while answering the buyer's question.
- Start before recognition. Leaving the name out lets us see whether an answer brings your company into the buyer's search.
- Read the shortlist. We look at the names an assistant recommends because buyers often make their choice from the companies they consider early.
- Compare like questions. The same wording goes to each engine, so we can compare how they answer the same buying need.
It fits startups that AI answers still skip
Act on grader findings
At a pre-seed to Series A B2B company, we can work through the grader findings that otherwise wait for a marketing hire.
Enter service shortlists
AI-enabled service companies can test the questions buyers research before they speak with a provider.
Investigate missing recommendations
Companies with well-built pages can find out which competitors assistants choose and examine the material behind those answers.
Talk to Tin
Let's discuss the questions your buyers ask assistants and the companies they hear about today. The page checks and answer tests run with no long-term commitment.
FAQ
Do you give us a score?
You get findings from the page checks and a record of which answers mention, cite or recommend your company. We show whom the answers name instead and the evidence behind the findings, so the next page change has a clear reason.
Which assistants do you ask?
We use ChatGPT, Perplexity and Gemini for the service check. We compare their replies to the same questions to see how each describes the market.
Should we fix our grader's list before we talk?
We can discuss the list as it stands. Checking the answers helps us choose which page fixes to address first, and making those fixes is part of the work we take on.
Can our own team run the same check?
Your team can use Tin's open-source AI visibility audit. It asks five buyer questions without naming your product and saves its findings as AI_VISIBILITY.md in your project. Our service check uses the broader set of buyer questions described above.
Choose the buyer questions worth testing first
Talk with the founders and engineers who run these experiments about the questions buyers ask in your market. We can discuss what the answers say today and which pages would be useful to examine first.