Key Takeaways
- Stop counting stars and start counting sentences. A five-star average is a number. An assistant answering a buyer needs words it can repeat, and an average carries none.
- Check where each piece of proof actually lives. Most coaching proof sits in messages, camera rolls, and expiring posts. None of that is reachable, however good it is.
- Assume nothing spoken counts until it is written. A video testimonial with no transcript is a warm asset for visitors and an empty one for anything reading the page as text.
- Match your proof to the question being asked. Nobody asks an assistant for a coach who is great. They ask for a coach for a specific person with a specific problem, and your proof has to say that.
- Fix the reachability before you chase more reviews. Doubling the number of reviews in a place nothing reads doubles nothing. Moving what you already have is the cheaper win.
This is the most frustrating version of the problem. The business is good. The clients are happy. The reviews say so, in public, with their names on them. And an assistant asked to recommend someone in your town names three people who are not you.
It feels like a judgement. It is not one. Nothing weighed your proof against a competitor's and preferred theirs, because in most cases your proof was never in the room.
That is the proof problem, and it is mechanical rather than mysterious. Here is where it happens, and what it takes to close it.
Why doesn't AI mention me when my reviews are great?
Because an answer is built from text a machine can reach, and most great proof is not reachable text. Praise sitting in messages, screenshots, video with no transcript, or a rating with no words behind it is invisible to the thing writing the answer.
Two separate things have to be true before a review can turn into a mention. It has to be somewhere public that can be read, and it has to contain something specific enough to be worth repeating. Great reviews routinely fail one of those and often both.
The good news in that sentence is the size of the job. You are not being asked to become a better coach or to go and earn new results. The results exist. They are in the wrong container.
Where the proof problem actually sits
In the gap between proof that persuades a person and proof that survives being read by a machine. Those are different tests, and almost everything a coaching business collects passes the first and fails the second.
A screenshot of a client's message is persuasive on a sales page and completely silent as evidence, because the words inside it are pixels. A before-and-after photo is the same. A voice note is the same. This is the same failure covered from the other direction in making client results legible to AI, and it accounts for more missing mentions than any ranking factor.
The second half of the gap is specificity. "Highly recommend, great coach" is real, public, readable, and still useless, because no buyer asks an assistant to find them a coach who is great. They ask for a coach who works with people like them, on the problem they have, in the place they are.
What makes a piece of proof reachable
Three properties, all required at once: it is public, it is text, and it is attached to your business name where anyone can see the connection. Proof missing any one of the three does not reach the answer.
- Messages from clients
- Screenshots of praise
- Video testimonials, no transcript
- Before and after photos
- Stories that expire
- Public reviews with detail
- Written testimonials on your page
- A case study anyone can open
- Public reviews with detail
- Written testimonials on your page
- A case study anyone can open
In words, because the figure holds no claim of its own. The eight common forms are messages from clients, screenshots of praise, video testimonials with no transcript, before and after photos, stories that expire, public reviews with detail, written testimonials on your own page, and a case study anyone can open. The last three pass. The first five do not, and they are usually where most of a coach's best evidence lives.
Notice what is not on the failing side: quality. Nothing there failed because it was weak. It failed because of where it was kept.
What a great review has to survive to be repeated
Five stages, and a review only counts if it clears all of them. Most stall at stage one or stage three, which is why the finished review can be excellent and the outcome can still be silence.
| Stage | What has to be true | Where it usually breaks | The fix |
|---|---|---|---|
| 1. Published | The words exist somewhere public | The praise stayed in a private message | Ask for it on a public platform, with permission |
| 2. Readable | The words are text, not an image or audio | Screenshots, photo quotes, untranscribed video | Type it out, or publish the transcript |
| 3. Specific | It names a person, a change, and a timeframe | "Great coach, highly recommend" | Ask three questions instead of asking for a review |
| 4. Attached | It is clearly connected to your business name | A first-name-only page nobody can match to you | Same business name, everywhere, spelled the same way |
| 5. Durable | It is still there next month | Stories, temporary posts, deleted highlights | Keep a permanent copy on a page you own |
Stage four is the quiet one. Plenty of coaches trade under one name on a review platform, another on their site, and a shortened version on social, and nothing joins them up. Google's own guidance on local business structured data exists partly to remove that ambiguity, and the underlying vocabulary for a review is the schema.org Review type. Both are useful later. Consistent naming in plain text is what matters first.
The order to fix it in
Move what exists before you collect anything new. The proof you already have is the cheapest material available, and most of it needs a copy and a permission, not a campaign.
- Write out the best five things clients have said to you, ask permission to publish each one, and put them on a page as plain text. An afternoon of work, and it is the highest-value afternoon in this list.
- Transcribe every video testimonial you have. The video stays where it is. The words go underneath it, where they can be read.
- Fix the naming. One business name, spelled identically on your site, your profiles, and anywhere you are listed.
- Then ask for new reviews properly. Three questions rather than a request, which is the whole method in asking for reviews AI can find and trust.
- Write one case study a buyer could check. Who they were, what changed, over how long, with which service.
Here is the honest part. Doing all five does not guarantee a mention. Reachable proof is a requirement, not a lever you pull for a result, and plenty of businesses need other things fixed as well, which is the wider list in why ChatGPT doesn't recommend your coaching business. What it does guarantee is that when something does read you, there is finally something to read.
Google documents how its own AI features use web content, and OpenAI publishes which crawlers it operates. Neither describes a scoring system you can game. Both describe systems that read published pages, which is exactly why the container your proof sits in decides so much.
Our free check asks twenty-five buyer questions for you at $0 and emails a plain-English readout within the hour. It runs as soon as you ask for it.
Questions we hear the most
My reviews are great, so why doesn't AI mention me?
Because an assistant repeats text it can reach, and most great proof is not reachable text. Praise in messages, screenshots, and untranscribed video is invisible to it, and a star average with no detail behind it gives it nothing to say about you.
Do AI assistants look at star ratings?
They read what is published in public, and a rating is a small part of that. The words in a review matter more, because words are what an answer is built from and a number cannot be quoted usefully.
I have hundreds of reviews. Is that enough on its own?
Not on its own. Volume without specifics reads as one repeated sentence, and it still does not answer the question a buyer asked. Ten reviews naming a starting point, a result, and a timeframe do more work than hundreds saying great coach.
Do testimonials on my own website count for anything?
Yes, if they are plain text in the page rather than images or a slider that loads afterwards. They carry less weight than an independent platform because you control them, but they are readable, and readable is the bar most proof fails.
Why does AI name a competitor with fewer reviews than me?
Usually because their proof is easier to read, not because it is better. A competitor with a plain page describing who they help and what changed is quotable, and quotable beats impressive when an assistant is assembling a short list.
How long does it take for a fix to show up in answers?
Weeks to months, and honestly it varies. Pages have to be found and re-read before anything you publish can be repeated, so treat this as a slow signal and measure it monthly rather than daily.
How do I find out whether AI is using my reviews at all?
Ask an assistant what people say about your business, then read the answer against what your clients actually wrote. Vague, wrong, or borrowed-from-somewhere-else means the proof it can reach is thin, stale, or in a place it cannot read.
What does the AnswerHalo free check include?
It costs $0. We ask ChatGPT 25 real buyer questions from your niche, then email a plain-English readout within the hour: how many answers named you, who was named instead, and the first fix. It runs as soon as you ask for it.
Your buyers are already asking. Find out what AI tells them.
The free check asks ChatGPT 25 real buyer questions about your niche and sends you the report within the hour, at $0. If it shows you are already getting named everywhere, we will say so in plain words.
Get my free check →