Service pages AI recommends: optimize so AI tools pick you
Customers no longer just type "spa district 7" into a search box. They ask an AI assistant a full question: "which spa in District 7 is good for sensitive skin?", "which event company in Ho Chi Minh City can I trust with a 300-guest conference?". The assistant answers with a shortlist of 3 to 5 names, each with a reason. Whoever is on that list gets the customer; whoever is not does not even get compared. This article is about exactly one job: optimizing your service page so AI tools choose to recommend you.
Quick summary: According to BrightLocal's Local Consumer Review Survey published in 2026, 45% of consumers used AI tools to find local business recommendations in the past 12 months, up from 6% a year earlier. When asked "best spa in District 7?", an AI tool does not scan the whole internet; it rereads a small set of sources it trusts: Google Business Profile, review sites, industry directories and the provider's own service page, then picks providers whose information is clear and consistent. To get picked, a service page needs 6 things: a direct answer at the top covering price, area and differentiator; Service plus LocalBusiness schema; NAP consistency with Google Business Profile; proof in numbers; an FAQ matching the questions customers ask AI; and mentions on the sources AI reads most.
What does AI base its provider recommendations on?
Asking AI "where should I book?" is no longer an early-adopter habit. BrightLocal's Local Consumer Review Survey 2026 found that 45% of consumers asked an AI tool for local business recommendations in the past 12 months, versus just 6% in the previous year's survey. Among 30 to 44 year olds, exactly the group that spends most on spas, restaurants and events, the figure reaches 64%. In the same survey, 42% of respondents said they trust recommendations from AI platforms as much as written customer reviews.
So what does the machine pick names from? Given a question like "trustworthy event company in Ho Chi Minh City", an AI tool typically switches on search and pulls back a small set of source pages: Google Business Profile listings, review sites and directories, local "top providers" articles, and the service pages of the providers themselves. It then synthesizes: providers mentioned across several sources with consistent information, with a clear description of which area they serve, roughly what they charge and what they are best at, make it into the recommendation list with a one-line reason. In other words, the competition does not happen when the customer reads your page. It happens earlier, when the machine reads it.
A note on scope: BrightLocal surveys US consumers. Vietnam has no study of the same scale yet, but the direction matches what we see in our clients' chat inboxes: more and more conversations open with "I asked an AI and it mentioned you".
A direct answer at the top: price, area, differentiator
Most Vietnamese service pages open with a slogan: "Elevating experiences, touching emotions". A machine reads that slogan and extracts nothing it can recommend. The highest-impact fix, at near-zero cost, is a direct-answer block right under the headline, 80 to 120 words, covering the four things both the customer and the machine need: what the service is, which area you serve, the price range, and your biggest differentiator.
An example for a spa in District 7: "An Spa specializes in facial care and herbal scalp treatments for sensitive skin, serving District 7 and Nha Be, Ho Chi Minh City. Single sessions run from 250,000 to 800,000 VND; monthly packages start at 1.8 million VND. What sets us apart: 100% of therapists hold a basic dermatology certificate, more than 4,200 client visits since 2019, and an average rating of 4.8/5 across 630 reviews." Those three sentences are what an AI tool quotes almost verbatim when it recommends you. This is also the core principle of GEO, explained in full in our article on what GEO is.
The GEO study by a Princeton University team and collaborators (published 2023, presented at the KDD 2024 conference), measured across roughly 10,000 queries, found that optimizations of this kind boost a page's visibility in generative engine answers by up to 40%; adding statistics, quotations and cited sources alone delivered 30 to 40% gains. Put simply: for the same service, the page written with concrete numbers gets named far more often than the page written with adjectives.
Service + LocalBusiness schema and consistent NAP
The direct-answer block serves the machine's "reading comprehension"; schema serves its "lookup". A service page should declare two layers of structured data: Service describing the offering (name, description, areaServed, price range) and LocalBusiness or Organization identifying the business (name, address, phone, opening hours, sameAs links to social profiles). An event company covering all of Ho Chi Minh City should fill areaServed with the districts it covers, or "Ho Chi Minh City", instead of leaving it blank.
Just as important is NAP consistency: Name, Address, Phone must match exactly across the service page, Google Business Profile, Facebook and every directory. Machines cross-check sources; a business showing 3 different phone numbers on 3 platforms reads as noise and is easily dropped from the shortlist. The Google Business Profile listing remains the strongest identity source for local queries; the detailed playbook is in our article on optimizing Google Business Profile in the AI era.
Customers ask an AI assistant on their phone before choosing a provider. Photo: Jernej Furman from Slovenia, Wikimedia Commons, licensed CC BY 2.0.
Concrete proof and FAQs that match what customers ask AI
Machines do not trust self-praise; they trust countable facts. Three kinds of proof belong directly on the service page, each with a number: scale of experience (years in business, number of projects or clients: "212 corporate events since 2018"), measured results ("68% repeat-customer rate", "4.8/5 satisfaction across 630 reviews"), and certifications, awards and flagship clients verifiable via links. An event company in Ho Chi Minh City writing "official sound and lighting partner of 2 major convention centers in District 1" gives the machine far more to quote than any slogan.
Next, the FAQ. A service page FAQ should not be written in internal jargon but in the exact words customers type into AI assistants: "how much does a sensitive-skin facial in District 7 cost?", "how far in advance should I book a 300-guest conference in Ho Chi Minh City?". Keep each answer 40 to 70 words, with numbers and area names, backed by FAQPage schema that matches verbatim. The detailed method is in our guide on how to write FAQs AI will cite.
Finally, remember the machine also reads pages that do not belong to you. In the same 2026 BrightLocal survey, AI became the number 3 discovery channel for local businesses, behind only Google and Facebook. Check whether the "top spas in District 7" and "best event companies in Ho Chi Minh City" round-ups that AI tools tend to cite include your name; one accurate mention on those pages is often worth more than many anonymous backlinks.
A 6-point checklist for service pages that want AI recommendations:
- Direct-answer block at the top: service, area, price range, differentiator, in 80 to 120 words.
- Service + LocalBusiness/Organization schema: complete areaServed, price range, address, sameAs.
- Consistent NAP: name, address and phone matching Google Business Profile and every directory.
- Proof in numbers: years, project counts, repeat rate, review score with review count.
- FAQ in the customers' words: 3 to 5 pairs of 40 to 70 words, with matching FAQPage schema.
- Presence on sources AI reads: review sites, industry directories and local round-ups naming you accurately.
Testing and measuring: how do you know you got picked?
The fastest check is to play the customer: open the AI assistants, ask the 5 to 10 questions your customers actually ask, and record which brands get named, whether you are among them, and which facts the machine uses to describe you. If it quotes the wrong price or the wrong area, some source almost certainly still carries outdated information. The full checking routine is in our article on checking your brand on ChatGPT.
On the technical side, confirm robots.txt is not blocking AI crawlers, the page carries published and modified dates, and the schema parses. All of these items can be checked automatically with the free Chạm AI SEO GEO audit tool. A sensible rhythm: re-test monthly, and after every price change update the service page, Google Business Profile and the directories within the same week so the sources never drift apart.
Frequently asked questions about service pages AI recommends
What does AI base its provider recommendations on?
AI tools reread the sources they trust: Google Business Profile, review sites, local media, industry directories and the provider's own service page. They favour providers whose information is clear and consistent across sources: what the service is, which area it serves, roughly what it costs, and what proof backs it up. Vague pages, or pages that contradict other sources, rarely get named.
What should you fix first on a service page so AI recommends it?
Add a direct-answer block at the very top: what the service is, which area you serve, the price range, your biggest differentiator, plus 1 or 2 proof numbers. That block is the part AI tools quote most easily. Only then move on to Service plus LocalBusiness schema, NAP consistency with your Google Business Profile, and an FAQ written in your customers' own words.
How do you know whether AI has already noticed your service page?
Play the customer: ask AI assistants the exact questions your customers ask, for example "best spa in District 7", then check whether your brand gets named and which facts the AI uses to describe you. In parallel, confirm robots.txt is not blocking AI crawlers and run the page through an SEO GEO audit tool to catch missing schema, TLDR and FAQ blocks.