Twelve questions, roughly in order of how much they tell you. The first four will sort most of the field on their own.
They work because they are easy to answer in a sentence if you do this work, and very hard to fake if you do not.
Why the usual signals do not help here
We measured 608 AI answers to the questions buyers ask when hiring in this category, across five engines over three months.
“We've always grown through referrals - builders who know us pass our name on. That works, but it only reaches people who already know someone in the industry. Within a couple of weeks of the content going live, we had someone contact us directly through the website. That's a channel we didn't have before - and the enquiries have kept coming.
Two things fall out of that. The category is concentrated, so being merely competent gets you nothing. And the sources that decide the answer are mostly not anybody's own website, which means most proposals you receive are aimed at the wrong half of the problem.
The questions below are built to find out whether the agency in front of you knows that.
The four that matter most
1. Can I see a prompt-level report from a real client?
The actual buyer questions, the engine each was asked on, and whether the client was named in the answer. Redacted is fine. Refused is not, because the report format is the thing you are assessing.
Bad answer: a traffic chart, or a single visibility percentage with no underlying prompts.
2. What is the difference between being cited and being recommended?
Being cited means your page was used as a source. Being recommended means your business was named in the answer. Different causes, opposite fixes, and a business can have a great deal of one and almost none of the other.
Across the answers we measured, the brands that supply the reading and the brands that get the recommendation overlap far less than anyone expects. Plenty of companies are read constantly and named almost never.
Good answer: they explain the distinction unprompted and tell you how they measure each separately.
3. Which citation sources matter in my category, and why those?
Good answer: specific and industry-particular. Insurer directories and licensing bodies in healthcare, established legal directories in law, review aggregates in home services. We hold citation-source data across 27 industries and the high-weight sources are different in every one.
Bad answer: Google Business Profile and "the major directories". Those are the ones everyone names, and in most categories they are not the ones carrying weight.
4. How will you show me movement in the first month?
Not results in the first month. Movement, from a measured baseline, so you are never guessing about whether anything is happening.
Good answer: a baseline in week one, then weekly deltas against it.
Bad answer: a first report at day 60.
The next four
5. How do you decide what to do first?
Every audit produces forty items and four of them matter. Ask how they choose, and listen for whether the answer is about your category or about their process.
6. What happens if the number does not move?
A good answer has a review point, a description of what would change, and an exit. A vague one means the contract term is doing the work.
7. Is your tooling yours, or licensed?
Either is fine. Licensing good software is often the right call. Evasion about which it is is the problem, because it is a small dishonesty about something easy to verify.
8. Who actually does the work?
In larger agencies, AI search often sits inside the SEO team as one line of a broader retainer, staffed by whoever is free. Worth knowing before you sign rather than after.
The last four
9. What is your view on community content?
In the category we measure, Reddit is cited 225 times across 608 answers, ahead of every agency website in the field. YouTube is cited 145 times and LinkedIn 134. An agency with no view on this has not looked at where the citations actually come from.
10. How do you verify claims about my competitors?
If they will write comparison content, ask where the facts come from. The correct answer involves the competitor's own published material and a date, not a third-party estimate. An agency careless about a checkable claim regarding a named company will be careless about a claim regarding yours.
11. What do you need from me each month?
Done-for-you only saves time if it genuinely is. An arrangement where they draft and you rewrite is slower than doing it yourself.
12. What does month one look like specifically?
A good answer is concrete and mostly unglamorous: measurement, a source audit, claiming. If month one is a strategy document, ask what happens in month two.
What the answers look like when the work is real
The proof of a method is what happened to the businesses that used it.
PROCERT had grown entirely on referrals for years. Within a couple of weeks of the work going live they took their first direct inquiry from someone who had never met them, through the website. Read the detail on PROCERT and Fur Magic.
What not to weight too heavily
Years in business. The field is three years old. Anyone claiming a decade of AI search experience is describing a different job.
Volume promises. Twenty articles a month is throughput, not an outcome. Ask what the reporting measures instead.
The size of the agency. It predicts predictability, not depth. In a field this young, whether the specific people on your account have looked at the data matters more than the size of the firm around them.
Do this before any meeting
Run the free test. Ten prompts your buyers would genuinely ask, in a logged-out session, two columns: named, and cited. Bring the result.
It changes the meeting entirely. You will know which of the two problems you have, and you will be able to tell within five minutes whether the person opposite recognizes the distinction.
What a good answer sounds like in practice
Question three, the one about citation sources, is the hardest to fake. Here is the same question answered two ways.
A weak answer: "We would start with a full audit of your online presence, make sure your Google Business Profile is optimized, and get you listed in all the major directories. Then we would build out authority through consistent content."
Everything there is true of every business in every category. Nothing in it required knowing what you do.
A strong answer: "For a building certifier in New South Wales the register entry does more work than anything on your site, because a model treats it as verification rather than a claim. After that it is your professional association listing and two review platforms your category actually uses. Google Business Profile matters and it will not differentiate you, because every competitor has claimed it."
That answer is specific, slightly surprising, and checkable in five minutes. It also tells you the person has looked at your category rather than at their own process.
Ask the same question of all three agencies on your shortlist and the difference is usually decisive on its own.
Ask us the twelve
We will answer all of them on a call, and we will bring your own numbers rather than a generic deck: your real buying questions, run across all five engines, showing which name you and which name a competitor.
Bring us the twelve questions
A free AI visibility audit on your own buying questions, across all five answer engines.
Book your audit
Ashur Homa
Built and scaled a digital brand to $100M+ in sales with zero ad spend. Has helped businesses generate millions through AI go-to-market strategy. Leads growth at Omni Eclipse.
Connect on LinkedIn