Ask an AI assistant "who's a good plumber near me?" and watch what happens. It doesn't return a page of ten blue links. It names one, maybe two or three businesses, in a clean little paragraph — and stops. For an agency, that single behavior change is the whole new game: your client is no longer competing for a spot on a page, they're competing for a seat on a very short list.
For twenty years, search was a list. A client could be result number seven and still get the click. The work was to climb the list, one position at a time. AI recommendation is not a list — it's a shortlist. The assistant has already done the comparing, filtering, and deciding on the customer's behalf, and it hands back a verdict, not options. There is no page two. There is barely a page one.
This is becoming the default fast, which is why it belongs in every client conversation. Today, 37% of people start their searches with AI, and for local recommendations specifically, usage jumped from 6% to 45% in a single year — a 650% increase. ChatGPT alone fields around a billion searches every week from roughly 800 million weekly users, and Gartner expects traditional search volume to fall 25% by the end of 2026. The shortlist is where the customers are going, and it has far fewer seats than the old results page ever did.
To sell this work, you have to understand the assistant's point of view. An AI that recommends a local business is not running a popularity contest. It is managing its own risk. Every time it names a plumber, a dentist, or a roofer, it is staking a little of its credibility on that answer being good. If it confidently recommends a business that turns out to be closed, unlicensed, or unreachable, the person asking learns not to trust it. The assistant's entire value depends on being right, so it behaves like a cautious referrer who only vouches for businesses it can actually stand behind.
That caution explains the short list. The assistant would rather name two businesses it is sure about than ten it is merely guessing at. It is looking for a few specific things before it will put its name next to yours: is this business real (does it clearly exist and operate), is it trustworthy (do others vouch for it), is it findable (can the AI actually read and confirm the details), and is it consistent (does the story hold together everywhere it looks). A business that answers all four cleanly is a safe recommendation. A business that leaves any of them ambiguous is a liability the assistant simply routes around.
So when there are only two or three seats, the math gets brutal. In our scans across local markets, only about 1.2% of local businesses — roughly 1 in 83 — get recommended by AI at all. The other 82 are not bad at their jobs. They are invisible because the assistant never got enough confidence to name them out loud, and an AI that isn't confident doesn't hedge — it just picks someone else.
Reframed this way, the job your client is paying for becomes clear: you are not "doing SEO" so much as making their business easy for a cautious machine to vouch for. The good news, and the reason this series exists, is that the things that build that confidence are not mysterious or manipulative. They are specific, observable, and fixable — an active, correctly categorized business profile; consistent name, address, and phone everywhere the business appears; a healthy stream of recent reviews; a website the AI can actually read; structured data that spells out who they are; and independent sources around the web that agree on the facts.
One idea is worth getting straight up front, because it reframes the entire list. Some of these factors the AI reads directly — the profile, the reviews, the structured data, the words on the page. Others it never reads at all, yet they still decide the outcome indirectly. A slow or confusing website, for example, never sits in front of the assistant — but it drives real customers off, and the customers who bounce are the ones who leave thin reviews, skip the follow-up call, or drop an offhand complaint on a site the owner can't control. The AI reads those. So every factor reaches the recommendation one way or another: some because the assistant sees them, and some because they shape the people whose opinions the assistant trusts. Nothing on this list is busywork.
We call these the AI Ranking Factors, and the rest of this series walks through them one at a time, always from the same angle: what is the AI worried about, and how does this factor put that worry to rest. None of them are tricks. There is no overnight switch and no honest agency can promise the top seat tomorrow. But each factor you fix is one more reason for the assistant to be confident, and confidence is what gets a business named.
The hardest part of selling this isn't the pitch — it's the middle. The businesses winning the shortlist right now aren't doing one heroic thing; they're doing a dozen small things consistently, so that no matter where the AI looks, the story checks out. That consistency takes weeks to compound, and in those weeks your client will ask the uncomfortable question: is any of this working?
This is the part of the engagement that quietly makes or breaks retention. Momentum toward a recommendation is real long before the AI is actually naming the client — the profile gets completed, the mismatches get cleaned up, the reviews start flowing — but that progress is invisible to a client who is only checking whether their name appears yet. Showing that movement before the payoff lands is its own discipline, and client-ready progress reporting that demonstrates the factors improving week over week is the quiet answer to "are we there yet." It is the difference between a client who renews on faith and one who can see the needle moving.
That is the entire game: understand what the AI is afraid of, fix the factors that calm it down, and keep the client seeing progress while the shortlist slowly rearranges in their favor. The rest of this series breaks down each factor in turn — what the AI is checking, why it matters, and how to put it in language a business owner will happily pay to fix.