How many clients already buy through AI, and what to do about it
More and more people start a purchase with a question to an AI assistant, not with search. That is a change of habit. For a local business it means a recommendation in the answer matters more than tenth place in the results. If your name is not in the answer, the buyer never sees you.
Below: what is changing in behavior, who gets hit hardest, what does not work, and where to start before competitors settle into the recommendations for good.
What is changing in behavior
Before: search, compare, call. Now: one answer, "here are two decent ones", a call to the first. Fewer tabs, and less chance to "catch up" with retargeting.
Voice speeds the shift. Alice in the car, a speaker at home, Yandex Neuro in the app, ChatGPT at work. Three entry points, one principle: your name is in the answer, or it is zero.
The market does not split "online and offline". A client asked an AI assistant in the parking lot and walked into someone else's clinic. You only learn that from an empty booking, not from site analytics.
Numbers, without magic
There is no public figure for "the AI share of the market" in Russia. The behavior shows up in the field: owners hear "someone recommended others" more often, and managers hear "I found it through an AI assistant". That is not a Telegram poll. It is calls and the CRM.
Yandex Metrica will not show the source "Alice said to call here". Yandex Direct will not show a loss in the organic "who to choose". You need a separate rhythm: GEO Score, three test queries, the gap with the leader.
A younger audience more often starts with ChatGPT. An older one starts with Alice and Yandex Neuro. Close one channel and you hand over half the niche without a fight. The shift does not ask what age you target in Yandex Direct.
B2B is in this too: "a contractor for an implementation" in Yandex Neuro and ChatGPT. That needs cases, timelines, and licenses in HTML, not "we are a reliable partner".
Every month without an AIO layer, a competitor stacks signals: FAQ, listings, mentions. Catching up in six months costs more than assembling the base now. That is how the signals accumulate.
A voice scenario in the car: "book a dentist nearby". Alice names one or two clinics. If you are not there, the client drives to the neighbor. No click on your site.
ChatGPT at work: "a turnkey website contractor in [city]". B2B is in this too. You need cases, timelines, and contacts in HTML, not a PDF of a presentation.
Parents' kids ask an AI assistant about clubs, clinics, and tutors. A family business with no FAQ loses the generation that does not scroll search results.
Aggregators and marketplaces in "where to buy" answers are the pattern for e-commerce. Your own store with no Product schema stays out of the answer, even with ads running.
How this differs from a fad
A fad passes. The habit of "ask an AI assistant before you buy" stays. If you are not in the answer, you are outside the path to a call. You are not "a bit behind".
Every month without an AIO layer, a competitor stacks signals. Catching up costs more than assembling the base now.
A redesign with no robots.txt, no FAQ, and no listings is a pretty storefront in a closed mall. The shift walks past it.
Who gets hit hardest
- Local services: repairs, clinics, salons, delivery;
- B2C where the choice is "who to trust";
- Niches with standard questions about price and time.
B2B with a long cycle is in this too: "a contractor for …" more and more often starts with an AI assistant. Waiting until "it settles" means handing the niche over.
What does not work
Ads alone, with no AIO layer. A blog about AI instead of a price list and an FAQ. Blocking bots out of fear. A redesign before the listings and technical access are lined up.
Five signs you are outside the shift
- Clients ask "are you in Alice?" and you do not know the answer.
- The SEO report is green. Yandex Neuro names the neighbor.
- You are waiting "until people get used to AI". Competitors are already in the answers.
- A redesign is the priority. Nobody has looked at robots.txt in a year.
- There is no GEO Score and no monthly test queries.
The shift does not ask whether you are ready. It takes the leads from whoever assembled the AIO layer.
What people do themselves, and where it breaks
They wait until "it settles"
Competitors stack signals. You lose leads quietly.
A blog about AI instead of a price list
Ten articles, prices in a PDF. The model does not quote the file.
A redesign first
A pretty storefront, and robots.txt was never touched. The shift walks past.
Ads only
The click is paid. Alice's answer is someone else's name. A double spend.
What works
GEO Score every month. Matching listings. An FAQ built for commerce. Open bots. A full audit once a quarter. Implementation from an action plan, not chaos. Leadsy runs six platforms in one cycle. More: how it works.
A forecast with no dates
The share of queries through AI assistants will grow until answers feel more ordinary than search. A business with no AIO stays in ads and word of mouth, while competitors take the "free" recommendations.
People ask an AI assistant because it is faster than scrolling results. The habit is already there. The question is whether your brand is in the answer.
A workable order
- A free audit and GEO Score.
- The critical fixes: bots, NAP, FAQ.
- Another snapshot in a month.
- A full audit, or done-for-you, if there is a gap with a competitor.
Where to start today
The shift does not wait until you finish the design. Visibility first, then the storefront. The base: AIO and Alice.
Assemble the AIO layer now. That is cheaper than closing a gap once half the niche is already in the recommendations. Audit first, then fixes.
Ads with no AIO are a double spend: the click is paid, and Alice's answer sends people to the business the model knows better. The shift does not cancel Yandex Direct. Without a layer for AI assistants, you pay to feed someone else's clients.
A short checklist: GEO Score once a month, three test queries, one phone everywhere, an FAQ from calls, bots open. Five items, and you are at least in the game.
Waiting "until clients get used to it" means handing the habit to competitors. They are already in the answers. You are still at "we'll do a redesign".
"We already have clients" and "an AI assistant recommends us" are different things. Word of mouth does not scale. A recommendation in the answer does.
Leadsy runs six platforms in one cycle. You do not need six contractors and six reports that do not match.
After the audit, three queries out loud for the niche. Three times, other names: you are not in the shift. You are off the path to the call.
A free snapshot does not force you to buy implementation. It does show whether you are handing the niche over right now. Guessing costs more than a plan.
A post in Yandex Business, a case on VC, a normal FAQ: places where the name is tied to the niche. One article beats ten press releases with the same text.
Who should own AIO in the company: the owner approves the facts, marketing owns the FAQ, the developer owns robots and schema. One freelancer "for the copy" will not close the chain.
Do not wait for a law or a "required button". Clients already ask AI assistants. The question is whether your brand is in the answer.
Questions: FAQ. Plans with done-for-you implementation: LeadsyLabs.
A short checklist: GEO Score, three test queries, one phone everywhere, an FAQ from calls, bots open, a repeat in a month. Six items, and you are in the game.
Ads with no AIO feed other businesses in the recommendations. The shift does not cancel Yandex Direct. Without a layer for AI assistants you pay for the click, and the answer sends the person to a competitor.
Typical case: "we'll redesign, then we'll deal with AI". Six months later the competitor is steady in Alice, and you have a new storefront and the old robots.txt.
One decent case on VC or Dzen beats ten identical press releases. The name is tied to the niche. The model sees that.
The free audit in the form below is a snapshot before any edits. Without numbers, any advice from a chat is a gamble with the domain.
A local business with no AIO in a year is a shop with no sign on a new street. People walk by and do not know your name. The AI assistant is already naming the neighbors.
A redesign with no robots, no FAQ, and no listings is a new storefront in a closed mall. The shift walks past while you pick a font.
A speaker, Yandex Neuro on the phone, ChatGPT at work: three doors, one principle. Your name is in the answer, or it is zero.
Marketplaces and aggregators in AI answers are a separate pain for e-commerce. Product schema and stock on the site, not only an llms.txt file.
Who runs AIO in the company: the owner owns the rhythm and the facts, marketing owns the FAQ and the listings, the developer owns robots and schema. One freelancer "for the copy" will not close the chain.
Tied articles: AIO, Alice, GEO Score. The shift is a reason to assemble the layer, not a reason to panic.
Typical case: "redesign first, AI later". Six months later the competitor is steady in Alice, and you have a new logo and the old Disallow. The order: audit, the critical items, then the storefront.
One case on VC or Dzen, with numbers, beats ten press releases. The name is tied to the niche. The model sees that on ChatGPT queries.
The shift does not cancel offline: a client heard the name in Alice and walked into the location. No name, they did not come. A sign on the door does not replace a recommendation in the answer.
"We'll wait for regulation." Competitors are not waiting. They stack FAQs and listings. You lose leads quietly while you discuss the design.
Six platforms in one Leadsy cycle is not six contractors. One report, one queue of fixes. The shift multiplies channels. A normal process keeps the work in one place.
A checklist for the quarter: GEO four times, three Alice tests every month, one gap with the leader, fixes on the red items.
The free snapshot in the form below is the first step. Without numbers, any advice about "the AI trend" is a gamble with the domain.