Why Word of Mouth Alone Won't Get You Into AI Answers, and What Does

For most coaches and consultants, word of mouth built the business. Someone had a good year, told a friend, and the friend called. That engine still runs, and it still works.

It just stopped being the whole story. A growing share of buyers now open an assistant first and ask it who to hire, and that assistant has never had a conversation with anyone about you.

This is what actually happens to a recommendation that was only ever spoken out loud, why it never reaches the answer, and the small amount of work that puts the same reputation somewhere a machine can read it.

Does word of mouth help AI recommend my business?

Only the part of it that ended up written down somewhere public. An assistant has no ears and no social circle. It works from text it can retrieve, so a referral that lived in a conversation left nothing behind for it to find.

This is not a judgment about the quality of your work. It is a plain limitation of how the tools are built. Google's own guidance on AI features in Search describes answers assembled from pages it has already crawled, and Common Crawl, one of the large public collections of web text that AI systems are trained on, is exactly that: pages. Not conversations, not phone calls, not the group chat where someone raved about you last Tuesday.

So the question is not whether your reputation is good. It is whether any of it exists in a form that can be read.

Where a spoken recommendation stops

It stops at the last person who heard it. A spoken referral moves one hop at a time, gets vaguer with each hop, and leaves no record behind, so nothing accumulates anywhere a machine could later find it.

That is fine when the next hop is a buyer. It is a dead end when the next hop is an assistant, because there is no hop to make. The chain simply ends, and the answer gets built from somewhere else entirely.

ONE RECOMMENDATION, TWO ROUTESILLUSTRATIVE

A client finishes work with you and is glad they hired you

Spoken only

1. A client tells a friend
2. The friend tells a colleague
3. The colleague half remembers it
4. Nothing was written down

Said in public, in writing

1. A client writes it as a review
2. The review sits on a public page
3. The page gets read and reread
4. Your name is available to repeat

Both routes start from the same happy client. The upper route is the one most established businesses are running, and it ends where it ends. The lower route is the same goodwill, said once in public and in writing, so it is still there to be read months later. The wording of each step is illustrative, written to show the shape rather than to report a result.

AnswerHalo answerhalo.com

CLICK TO ENLARGE

Notice what the lower route does not require: no new clients, no better results, no marketing budget. It is the same event, recorded once. That is why this is usually a short job rather than a long campaign.

What kinds of word of mouth an assistant can read

Public, written, and specific. Those three conditions decide it, and a recommendation that misses any one of them may as well not exist as far as an assistant is concerned.

Public rules out anything behind a login, which is where a lot of the best recommendations now happen. Written rules out speech, voice notes, and video that nobody transcribed. Specific is the one people skip: a rating with no words attached tells a machine that someone was satisfied, but not what you did, who you did it for, or what problem it solved, so there is nothing in it worth quoting.

The bottom three rows are the whole job. If you want the longer version of how to ask for the fourth one without it feeling awkward, that is in how to ask for reviews AI can find and trust.

Why the two tracks stopped overlapping

They used to feed each other, and now they run in parallel. A referral used to end with the buyer searching your name and landing on your site, which put your pages into the loop. Increasingly the buyer asks an assistant instead, and your pages are only in the loop if the assistant already knew about them.

There is a second, quieter change underneath that. When a buyer does search, a growing share of those sessions end inside the answer rather than on anyone's website, which means the referral no longer reliably delivers a visit even when it works. What that looks like in the numbers is in why ChatGPT does not recommend your coaching business.

Perplexity's help center describes an approach built on retrieving and citing current sources, which is a useful thing to sit with for a moment. The assistant is not consulting a reputation. It is consulting a reading list. Being well regarded and being on the reading list are two different achievements, and only one of them is something you can arrange.

How to give your reputation a written form

Take the recommendations you already earn and get three of them written down in public this month, in the words your buyers use. That is the entire first move, and it costs nothing but a few awkward asks.

Start with the clients who already told you something specific. Reply to them, quote the sentence back, and ask whether they would put that in a public review. Most people say yes, and the ones who say yes usually write a better review than a cold request would ever produce, because you have already reminded them what they thought.

Then write one page that says plainly what you do, who for, and what changes. Not a brochure. A page that answers one buyer question completely, in their vocabulary. A recommendation gives an assistant a reason to trust you, and a page gives it something to actually say about you, and it needs both before your name shows up in an answer.

Finally, keep referring and keep being referred. The two tracks work best together, which is the argument made at more length in are referrals dying. Nothing about this suggests you should spend less time on relationships. It suggests you should leave a paper trail behind them.

If you want to see where you stand before you start, our free check asks ChatGPT 25 real buyer questions from your niche at $0 and emails a plain-English readout within the hour, including who is getting named in the answers you care about. It runs as soon as you ask for it.