When AI recommends a business, how does it pick?

A growing share of buying now starts with a question to an AI, not a search box. In a 2025 Adobe survey, 53% of US consumers said they had used AI to research a purchase, and Adobe's analysis of more than a trillion retail visits found AI-driven traffic up several times over during the 2025 holiday season, converting roughly a third better than other channels. ChatGPT, Perplexity, and Google's AI Mode are all racing to answer "what should I buy" and "who should I hire" directly. The question for a small business is simple. When the AI recommends someone, how does it pick, and how do you get named?

What changed

All three big assistants turned shopping and local recommendation into a core feature in the last eighteen months.

• ChatGPT added shopping research in 2025, with sourced product details and side-by-side comparisons. It briefly tested its own in-app checkout, then in March 2026 stepped back to sending people to the merchant's own site to buy. (CNBC reported the pivot.) The discovery stayed; the checkout moved back to you.
• Perplexity built a shopping flow with in-chat checkout and a free merchant program that lets retailers share catalogs, specs, and reviews. It leans heavily on reviews when it answers.
• Google AI Mode combines Gemini with Google's Shopping Graph of tens of billions of product listings, and in January 2026 Google announced new Merchant Center attributes and a business-chat agent aimed squarely at conversational queries. (Google's announcement has the detail.)

The common thread: these systems answer from structured data and trusted third-party sources. They do not browse your beautiful homepage and form an opinion. They assemble an answer from feeds, profiles, reviews, and mentions.

Why it matters

This is a different game from chasing a blue-link ranking, and it rewards different things.

First, the traffic is small but high quality. Most analytics setups do not even tag AI referrals yet, so a lot of owners cannot see it. But the people who arrive after an AI pointed them your way show up further along, already half-sold, which is why they convert better. Treat the channel as real even while it is small.

Second, being recommended depends on being legible to a machine and being mentioned by others. Two signals do most of the work.

• Your data feeds. If you sell products and you are not in Google Merchant Center, or not in Perplexity's merchant program, you are largely invisible to their shopping answers. The feed is the on-ramp. No feed, no mention.
• Your reviews and citations. Perplexity references reviews in close to every product recommendation, and assistants lean on third-party sources they trust: Google reviews and local press for local businesses, sites like G2, Capterra, and Reddit for software. (Search Engine Land covers the schema and local angle.) A business with fifteen stale reviews is easy for the model to pass over in favor of one with two hundred fresh ones.

The uncomfortable part for a small brand is the same lesson from our AI Overviews briefing: recognition is the currency. The AI quotes sources it has seen mentioned, reviewed, and structured. If the web barely mentions you, the model has little reason to put your name in its answer.

Your move

You do not need to chase every platform. You need to be clean, fed, and reviewed. Five moves, in order.

1. Get into the feeds. If you sell products, set up a Google Merchant Center feed and enroll in Perplexity's free merchant program. This is the single biggest lever for product recommendations, and it is plumbing, not marketing.
2. Fix your Google Business Profile. For anything local, AI surfaces weight a complete, fresh profile: right hours, real photos, current services, and recent reviews. Update it like it is a storefront window, because to the AI it is.
3. Make reviews a habit, not a campaign. Ask every happy customer for a Google review, every time. Reviews feed both local AI answers and the Map pack, and they are the cheapest recommendation signal you own.
4. Add the basic schema. LocalBusiness, Product, Review, and FAQ schema markup help the systems parse and trust your pages. If your site runs on a mainstream platform, a plugin handles most of it.
5. Get mentioned in the places AI reads. A few honest third-party listings, a comparison roundup, a Reddit thread where real people vouch for you. You cannot fake this, and that is exactly why the AI trusts it.

Tools and resources

The work splits in two: be findable, then catch the visit when it comes.

• For being findable, the feeds above are free and matter most. Schema is a plugin on most platforms. Reviews are a habit you can systematize with whatever CRM you already use.
• For catching the visit, remember the checkout is moving back to your site. An AI can send a ready buyer to your page, but if that page has no clear next step they bounce. A booking and follow-up tool such as GoHighLevel keeps the lead from slipping, and our full review covers where it helps and where it annoys. Pair it with the capture mindset in our Meta Advantage+ briefing, where the same logic applies: scarce, valuable traffic, so waste nothing.

For the wider shift in how customers find businesses, our AI search hub tracks it as it moves.

Worth watching

Two things will decide how big this gets.

One is where the checkout ends up. OpenAI tried owning it, then handed it back to merchants. Google and Perplexity are still pushing in-chat purchase. If buying inside the assistant becomes normal, the assistant becomes the storefront, and your product feed becomes more important than your website. If it stays on your site, your job is to be the recommendation that earns the click.

The other is measurement. Right now most owners cannot see AI traffic in their analytics, which makes it easy to ignore. As tracking catches up and the numbers become visible, expect more competition for these answers. Getting your feeds and reviews right while it is still quiet is the cheap version of this work.