How to Rank in AI Search: What We Found Auditing 11,534 Businesses

In February 2026 we scored 11,267 South Carolina businesses on how ready they are for AI-driven search. Not one earned an A. Three percent earned a B. Sixty-three percent landed in the C band, and a third scored D or F.

The same businesses did fine on the old measure. Forty-eight percent scored an A on Google Business Profile completeness. Their average profile score was 80.6 out of 100. Their average AI readiness score was 56.0. Ninety-one percent of them had a gap of twenty points or more between what a person sees and what a machine can read.

That gap is the story, and it is closing on its own timetable whether or not anyone acts on it.

Why the gap exists

A Google Business Profile was built for a person to look at. It rewards a good photo, a decent star rating, a phone number that works. Most owners have done that much, because the payoff is visible and the interface asks for it.

AI search reads something else. When someone asks an assistant for the best plumber in their town, the model is not looking at your photos. It is assembling an answer from structured facts: your categories, your hours, your price level, whether your business has a clear identity it can resolve, whether anyone has written anything about you recently, whether the details agree across the places it checks. Those fields are unglamorous, several of them are optional, and nothing in the interface tells you they now decide whether you get named.

So businesses optimised for the thing they could see, and the thing they could not see became the one that mattered.

What AI search actually rewards

We scored seven signals. The weights are published with the study, and they are worth knowing because they tell you where the effort goes.

Structured data carries the most at 25 percent: hours, price level, and multiple accurate categories rather than one vague one. Trust signals are 20 percent, which is rating quality combined with review volume. Entity clarity is another 20 percent, meaning whether a system can tell what your business is and confirm it against a website. Freshness is 15 percent, and it is mostly about recent review activity rather than posts. Visual content is 10 percent. Owner responses to reviews are 5 percent. Accessibility information is the last 5 percent.

Read that list again and notice what is missing. There is no keyword to stuff and no page to write. Six of the seven are facts about your business that either exist in a structured field or do not.

The categories that are furthest behind

The spread by industry is wider than we expected. Pediatric practices averaged 45.8 across 1,065 of them, the worst score of any category in the state. General contractors came in at 49.7, cleaning services at 50.6, pest control at 50.7, and attorneys at 52.0 across 1,074 firms.

At the other end, roofing contractors averaged 63.0, urgent care 61.8, gyms 61.7, dentists 61.6, and plumbers 61.1. Every one of those is still a C.

The pattern is not about sophistication or budget. Law firms are not short of money. What separates the stronger categories is that they tend to be emergency or near-emergency services, where the pressure to keep hours accurate and respond to reviews is constant. Emergency services in our sample averaged 58.9 against 55.2 for everyone else. Operational discipline, not marketing spend, is what shows up in the score.

The test worth running today

Open ChatGPT or Google's AI Overviews and ask the question your customer would ask. Not your business name. The problem, plus your town. "Best pediatric dentist in Greenville." "Emergency plumber near Mount Pleasant." Whatever a stranger with your problem would type.

You will get two or three businesses named. Note who they are, because that is your competitive set now, and it is often not the same list you would have written yourself. Then ask a follow-up: what makes them good. The reasons the model gives you are the signals it found, and they will point at exactly what your own profile is missing.

Do this once a month. It takes four minutes and it is the cheapest competitive intelligence available to a small business right now.

What to fix, in order

Start with categories, because they are free and they are the single highest-weighted thing you control. Most businesses pick one primary category and stop. Add every secondary category that accurately describes what you do. Inaccurate ones will hurt you, so this is not an invitation to pad the list.

Then hours, including holiday hours, which almost nobody maintains and which is a freshness signal as well as a structured one. Then price level, which feels like exposing something private and is one of the fields models lean on hardest when a customer asks for an affordable option.

Then reviews, and specifically the rate rather than the total. Two to five new reviews a month with owner responses beats two hundred reviews from three years ago, because recency is a signal and a stale profile reads as a business that might not be operating.

Then the site itself. The profile is what gets checked; the site is what confirms it. If the two disagree about your hours or your services, the disagreement is the problem, not either version.

Why this window closes

Right now the bar is low enough that doing the fundamentals properly puts you ahead of almost everyone in your category. Zero A grades in 11,267 businesses is not a competitive market. It is an empty one.

That will not hold. The businesses that fix this in the next year will be the ones AI systems have been recommending for a year by the time everyone else notices, and recommendation histories compound the same way rankings do. The cost of starting later is not the work, which is the same work. It is the head start you hand to whoever moved first in your category and your county.

The full study, including the methodology, the county breakdowns, and a published table of everything the data cannot tell you, is at research.louislynnco.com. Every business we scored is searchable by name.