Your buyer asks an assistant which vendor to use. It answers with two or three names. I measure how often yours is one of them, then do the work to change it.
No call required. Invoice within 24 hours, fixed scope, 14-day unconditional refund, and you keep the document either way.
What’s the best help desk software for a support team of about 15 agents?
Eight questions a support manager actually asks, run across three assistants. The names above are what came back. A well-funded vendor in this category was named in 2 of the 24 answers, and I am not going to name them on a public page.
When someone asks an assistant for a vendor, they do not get a list to compare. They get an answer, with two or three names. If yours is not there, you did not lose a ranking position. You were not in the conversation.
Nobody who cannot do the work publishes how they measure it, so here it is. A fixed set of 20 buyer questions, agreed on day one and frozen. Four assistants. Fresh sessions, no memory, no custom instructions. Re-run weekly.
One question on its own is noisy: assistants rotate the order and the tail of the brands they name, so a single answer proves nothing. The aggregate of 20 questions across four assistants carries a margin of about 7 points, and averaging the last four weekly runs brings it near 3.5. That is why the set is 20 questions and not 5, and it is the reason I will not promise a move smaller than 10 points: below that, it is indistinguishable from noise.
Not a diagram of the method, the method itself: every question, every verbatim answer, model versions and dates from the July 29 run are in this file. One measured vendor is a prospect and stays masked; everything else is untouched.
Below is a real run from July 29, 2026: 40 buyer questions, three assistants, 120 answers. Competitor names are masked because it is a live category, but the numbers are untouched. Yours arrives with the names visible and the full answer text per question.
| Brand | Named | Avg. pos | Domain cites |
|---|---|---|---|
| Competitor A | 25 | 3.1 | 37 |
| Your brand | 18 | 4.7 | 23 |
| Competitor B | 18 | 4.7 | 19 |
| Competitor C | 15 | 3.3 | 23 |
| Competitor D | 13 | 5.2 | 11 |
Read it like this: you are tied on mentions with Competitor B, but sitting at 4.7 in average position against 3.1 for the leader. And your domain is cited 23 times, nearly as often as Competitor C, so assistants are reading you and still naming someone else. That gap is the work.
This is the half nobody works on. A tool that publishes to your blog cannot touch the official platform, a competitor’s domain, or YouTube, and in this category those three carry more weight than your own site.
The full document also carries every question with its verbatim answer per assistant, the model version and date, your baseline with its margin of error, and the list of questions where you are absent. That last list is the 90-day plan.
If volume already ran and the answer still names someone else, that is what this is for. If it has not, buy the cheap tool first. I would rather say that than sell you the wrong thing.
Tell me your domain and your category and I run eight buyer questions across three assistants, then send you what came back: whether you were named, and which competitors were.
Not traffic. Not revenue. Not position. Not any number past 90 days.
My first AEO engagement has not completed 90 days, so I have no data to put a percentage on, and I am not going to borrow someone else’s. When I have it, this section changes and the guarantee below gets a number.
If the diagnostic is not worth what you paid, email me within 14 days and I refund it in full. No form, no justification, no clawback on anything I already gave you. You keep the baseline document and every prompt in it.
On day one we agree on 20 buyer questions and record what four assistants answer today: full response text, model version, date, in a document you own. Within 30 days of the first published change your brand is named in at least one of those 20 answers, or the next month is free and I keep working until it is.
How it is measured: the same 20 prompts, the same four assistants, fresh sessions with no memory and no custom instructions. Named means your brand name appears in the answer text. You can re-run every prompt yourself.
Is one out of 20 noise? On its own, yes — single answers swing, and I am not going to pretend otherwise. This guarantee is not a statistic, it is my month on the line until your brand shows up. When the first engagement completes 90 days, this bar gets a number too.
Order it without talking to anyone. Invoice within 24 hours, fixed scope, 14-day unconditional refund.
The work, run async. Starts after a diagnostic, which is credited in full against your first month.
For when answer visibility is a sales channel and not an experiment.
Two retainers a month, and that is not a marketing number. One person writes the questions, runs the measurement and does the work, so the month has room for two new accounts and no more. Diagnostics are not capped. If the month is full I tell you the date I open up instead of putting you in a queue you cannot see. While a month has an open slot, the rate is what you see on this page. When it fills, the next opening starts higher.
Prices in USD, month to month, no lock-in on the retainers. The diagnostic is credited in full against your first month if you continue. The diagnostic tells you whether you need a retainer at all, and sometimes the answer is that you do not.
Four fields and you are done. I send the invoice within 24 hours, the baseline run starts when it clears, and the document is with you inside five business days.
Santiago Moya · Santiago, ChileI am Santiago, I work from Chile, and I run the measurement and the work myself. There is no team, no account manager and no junior writing your content.
What I can prove today is adjacent, and I would rather label it than dress it up: I took an online store I operate from almost no organic traffic to 6,270 monthly clicks in six months, with 98% of its revenue coming from organic and zero ad spend in the last measured period. That is SEO and conversion work, not AEO.
AEO is new for everyone selling it, including the shops charging $3,000 a month. The difference is that I measure it, I publish the method, and I will show you the number before you commit to anything.
Check me outside this page: LinkedIn, X and GitHub. Same name, same face, same person you will be emailing with.
I am taking on my first retainer clients at these rates, and one person doing the work means two new retainers a month. Those two things are related: the price is what it is because I am building the case studies, and the capacity is what it is because nobody else touches your account.
What you get for being early: my full attention, one of the two retainer slots, and a say in what the reporting looks like.
No, and the difference is measurable. Ranking first on Google does not put you in the answer: assistants build answers from sources they trust, and plenty of those are not your site. In the category I measured on July 29, the second most cited source was reddit.com. No amount of on-page work moves that.
If your problem is publishing volume, a $39 tool solves it and you should buy it. Those tools publish to your site and report Google Search Console clicks. They never check the answer. This starts and ends with the answer, and about half of what shapes it lives on domains you do not own.
A fixed set of 20 buyer questions, four assistants, fresh sessions with no memory and no custom instructions, re-run weekly. One question in isolation is noisy because assistants rotate the order and the tail of brands they name. The aggregate of 20 questions across four assistants has a margin of about 7 points, which is why the question set is 20 and not 5.
Because it is not a promise, it is a document, and you can see the format before you buy it: the sample on this page is the real deliverable with a real company's numbers in it. You get your own version of that, the full answers, your competitors, the work list, and a 14-day unconditional refund. If it is not worth what you paid, you email me and keep the document.
You can, and for eight questions once you should. It stops working when you need it repeatable: one question asked once tells you nothing, because assistants rotate which brands they name and in what order. You need a frozen question set, fresh sessions with no memory, the same run every week, and enough volume that the aggregate is stable rather than noise. That is a measurement system, not an afternoon. And the harder half is not measuring, it is getting cited on domains you do not control.
Then results move for reasons neither of us controls, and I will tell you that with the data instead of taking credit or hiding it. It is the structural risk of this channel and I would rather you hear it from me before you pay than discover it at month three.
No, and that is deliberate. This runs async: you fill four fields, I invoice you within 24 hours, the baseline arrives as a document, and every month you get a report plus a recorded walkthrough you can watch at 11pm and forward to whoever signs the invoice. Retainers need a few things agreed in writing before they start, and that happens over email. If you want a live call I will do one, but nothing here is designed around getting you on the phone.
The diagnostic gives you your number, the full answers from four assistants, every competitor that appeared and the work list. 14-day unconditional refund, and the document is yours either way.