A small web development agency started offering SEO in its home city. Within two weeks, ChatGPT was recommending it as an SEO agency in that area, next to established agencies.
The agency had published a couple of articles, some on its own site and some elsewhere. That was the whole campaign. One person with a laptop showed up in the same answer as agencies with 50 employees.
Disclosure: The agency is an Airefs customer, and we build Airefs. We have an obvious interest in this story, so we have kept the claims narrow and sourced the rest.
This is one case, not a controlled experiment. But it fits a pattern in the data. That pattern matters for anyone who sells size as a reason to hire them.
Three trends meet here. One-person businesses are booming. AI tools let small teams do big-team work. And AI answers are easy to influence when the competition is thin.
What happened in two weeks
The agency already built websites for local clients. It decided to add SEO as a service in the same area. Then it did what any small agency can do in an afternoon.
It wrote a few articles about SEO for businesses in its city. Some went on its own site, others on third-party sites. Days later, ChatGPT began naming it when asked for an SEO agency in that area.
Nothing about this required a large budget. There was no press campaign, no award and no years of accumulated authority. The agency simply gave ChatGPT something to find.
That last point is the one to hold on to. AI assistants do not rank agencies by headcount or revenue. They assemble an answer from whatever evidence they can retrieve. A new entrant that supplies evidence can enter the answer.
Why thin markets are easy to enter
A national query like “best SEO agency” is crowded. Hundreds of agencies publish about it, and thousands of pages compete to be cited. Breaking in takes real authority.
A query like “SEO agency in [small city]” is different. Few agencies publish anything specific to that place. The pool of pages that mention both the service and the city is tiny.
When the pool is that small, a handful of new pages can change the answer. ChatGPT has to fill a list, and it fills it with the businesses it can verify. If your agency has a page naming the service and the city, you are a candidate.
AI answers are built from searches, not rankings
One analyst logged over 200 ChatGPT sessions across 10 industries and 10 states. ChatGPT ran around 30 searches per local recommendation, read about seven distinct sources and named roughly eight businesses. The same analyst found that 80% of recommended businesses rank nowhere in the top 50 of Google or Bing.
Treat that as one researcher’s self-reported work, not a peer-reviewed study. It still points the same way as larger datasets. The answer is assembled from many small searches, so ranking first in one of them is not required.
Google rankings carry less weight than expected
Suff Digital compared 17,464 ChatGPT local recommendations across 79 US cities with Google Maps results. 88.4% of the recommended businesses were not in the Maps top three. 17.5% were not in the top 100 at all.
ChatGPT did name the Maps number-one business in 45.9% of city and service searches. It then filled the rest of its list from further down. A top Google position helps, but it does not close the door on everyone else.
For a small agency, this is the useful part. You do not need to win the Google fight first. You need to be present in the sources ChatGPT reads.
Smaller cities overlap more, but the lists stay open
The same study found that in 29 smaller cities, 20.1% of ChatGPT’s picks were in the map pack. In the 50 large cities, the figure was 7.6%. With fewer businesses per service, ChatGPT and Google land on the same names more often.
That might sound like bad news for a newcomer. It is not. Fewer incumbents means fewer pages to beat, and ChatGPT’s list still holds more names than Google’s three-pack.
Suff Digital’s separate analysis of 3,160 answers found that 95.1% of local answers include a list of business cards. The list holds 11.8 businesses on average. Google’s map pack shows three.
A list of about a dozen names has room for a newcomer. A list of three does not.
The one-person business boom
The agency in this story is not an outlier in its structure. Businesses with no employees are the largest group in the US economy, and they are still growing.
The Census Bureau counted 30,427,808 nonemployer establishments in 2023, up from 29,811,495 in 2022. That is about 616,000 more in a year. These businesses made up 78.4% of all US establishments and brought in nearly $1.8 trillion in revenue.
Startups show the same shift. Carta reports that solo founders now start 36.3% of new startups, up from 23.7% in 2019. Its 2026 ownership report puts the 2025 figure at about 36%, up from 31% in 2024.
Carta links the trend partly to AI. In its words, AI has expanded what individuals can accomplish in a finite amount of time. That makes it more feasible for one founder to both build and sell.
What this means for agencies
Every one of those solo businesses is a potential competitor to a mid-sized agency. Many are former agency employees who left and kept their clients. Others are freelancers who now deliver more than they could three years ago.
A buyer used to compare an agency’s size against a freelancer’s. A buyer now asks ChatGPT, and the answer does not mention size at all.
This is the structural change. Discovery moved from sales calls and Google results pages to AI answers. The signals that favored large firms lost some weight. A large team, a glossy office and a long client list only count if they appear in the sources.
AI lets one person do the work of a team
The second trend is the one that makes the first one viable. A single operator can now research, draft, edit and publish at a pace that once needed a content team.
Consider what the agency in this story did. It chose a topic and wrote a few articles. Five years ago, that would have taken a writer, an editor and a part-time outreach person. Today one person can do it between client calls.
AI also lowers the cost of the surrounding work. A solo founder can produce proposals, reports, landing pages and outreach emails without hiring for each. The agency’s overhead stays small, so its prices can stay competitive.
Output is not the moat
There is a trap here. If every small agency can publish more, then publishing more is not an advantage. The advantage comes from publishing the right pages in the right places.
The Cheers Market Baseline analyzed 9,434 ChatGPT answers about home services in 100 US and Canadian metros. ChatGPT cited Expertise.com in 32.1% of answers and Angi or HomeAdvisor in 29.4%. Of the answers that cited anything, 84% cited at least one directory, review platform or “best of” list.
That dataset covers home services, not marketing agencies. Directories for agencies differ. But the lesson transfers: the pages that decide who gets named are often pages the business did not write.
A small agency that understands this places its articles and listings on the sources AI already trusts. A large agency that only publishes on its own blog may never appear.
Why size stops working as a moat
Large agencies built their position on signals that humans respect. Case studies, awards, headcount and press mentions all help a buyer feel safe. Those signals still matter once a buyer reaches the shortlist.
The problem is getting onto the shortlist. For local queries, that step is now partly decided by an AI model reading a thin set of sources. In that step, the model weighs evidence, not reputation.
Big agencies publish for a national audience
Most large agencies write about their services in general terms. Their service pages describe SEO, not SEO for dentists in a particular city. They have offices in a few places and content for none of the others.
That leaves a gap in every city they do not serve directly. A local agency that writes specifically about the service and the place fills the gap. ChatGPT, asked about that place, finds the local page and not the generic one.
Evidence can be built fast
A reputation takes years to build. A page that states what you do, where you do it and for whom takes an afternoon. In a thin market, the second can be enough to enter the answer.
This does not mean reputation is worthless. It means that for this one step, getting into the candidate list, reputation is not the only route. A newcomer has a path that did not exist when discovery ran through Google rankings alone.
What this does not prove
An honest read of this story needs limits. We should state them plainly.
It is one case. We watched one agency appear in one city for one type of query. Other agencies may need longer, or may not get in at all.
AI answers are unstable. Cheers asked four AI engines the same local question in the same week. 77% of the businesses named were named by only one engine. Appearing in ChatGPT does not mean appearing in Gemini, Perplexity or Google’s AI Mode.
Early gains can reverse. A thin market is easy to enter because few competitors are trying. Once they notice, they can publish too.
Blog posts are not the whole story. In the Cheers home-services data, contractor homepages made up 37.4% of cited pages and blog posts only 1.2%. Service pages and listings matter as much as articles.
Some queries stay hard. A one-person agency is unlikely to win “best SEO agency” outright. The opening is in specific places, niches and use cases.
What a small agency should do
The playbook is short. None of it requires a budget. It does require consistency, and a willingness to be specific.
1. Pick a place and a service
Do not target “SEO agency.” Target “SEO agency for [industry] in [city].” The narrower the query, the thinner the competition.
Write down five to ten queries a local buyer would ask. Include the city name, the service and a qualifier such as the industry or company size. These become your tracking list.
2. Publish pages that name the service, the place and the proof
Create a service page for each combination you want to win. State what you offer, who it is for and where you work. Add real results, client names where permitted, and pricing ranges if you can.
Then write articles that answer the questions buyers ask before they hire. Our guide on how to rank in ChatGPT covers the content formats that get cited. Keep each article specific to the place.
3. Publish off your site as well
The agency in this story placed articles on and off its own site. That is likely not a coincidence. AI models weigh third-party mentions because they are harder to fake than your own claims.
Look for local business blogs, chamber of commerce sites, industry newsletters and guest-post opportunities. Each mention that names you, your service and your city adds evidence. Our data on which sources AI search cites shows how heavily third-party pages dominate.
4. Get listed where AI looks
Check which directories and review sites show up for your category and city. For local services, Suff Digital found Expertise.com in 28.4% of ChatGPT’s local answers, Birdeye in 24.5% and Angi in 21%. Agency categories use different directories, so check your own results.
Keep your name, address, phone number and description identical across every listing. Ask recent clients for reviews on the platforms that matter.
5. Join the conversations AI already cites
Reddit and similar forums feed AI answers, including local ones. A short, useful reply from an account with real history can put your name next to the question. We cover the approach in our piece on using Reddit to influence AI answers.
Do not spam. A reply that helps the asker and mentions your agency once, honestly, lasts. A promotional drop gets removed.
6. Measure what AI says
Publishing without measuring is guessing. You need to know whether ChatGPT names you, in which position and for which prompts. You also need to see which competitors appear and what sources ChatGPT cites for them.
This is the part most small agencies skip. It is also where a small agency learns the most. A thin market changes slowly enough to show cause and effect.
What a large agency should do
Large agencies are not doomed. They have resources a solo operator lacks, and they can use them well. The first step is to stop assuming reputation is visible to AI.
Check your presence in every city where you have an office, and in the cities where your clients are. Ask ChatGPT the questions your prospects ask. Note which competitors appear, and which sources the model cites for them.
Then close the gaps. Build location and industry pages that go beyond a generic service description. Make sure your case studies are published on third-party sites, not just your own.
Most of all, treat AI visibility as a separate channel with its own measurement. Our case study on Growth Division shows a deliberate program in a crowded category.
Questions every agency owner should answer
The original post ended with two questions. They are still the right ones.
Do you know if your clients show up in ChatGPT for their city? If you run SEO for local businesses, this is now part of the job. A client who ranks well on Google but is absent from ChatGPT loses buyers you cannot see in reports.
Do you know which agencies show up for yours? Your competitors may no longer be the firms you meet at conferences. They may be one-person shops you have never heard of, named in answers you never checked.
You can run a first check by hand. Open ChatGPT, ask for the best agency in your service and city, and record who appears. Repeat with five prompts, then repeat a week later.
Our guide to setting up AI visibility tracking in one hour walks through the full setup.
Frequently asked questions
Can a one-person agency really appear in ChatGPT next to larger agencies? Yes, in thin markets. The case in this article took about two weeks. Results depend on how many competitors have published specific content about your service and place.
Is this only true for local queries? It is most true for local and niche queries. National, category-level queries are crowded, and winning them takes authority that a new entrant lacks. The practical opening is a specific place, industry or use case.
Does ranking on Google still matter? Yes. Suff Digital found ChatGPT names the Google Maps number-one business in 45.9% of local searches. But 88.4% of its picks fall outside the map pack, so Google rankings do not decide the whole answer.
How long does it take to appear in AI answers? It varies. The agency in this article appeared within two weeks, which is faster than most. Plan for weeks to months, and track progress rather than assuming.
Will this advantage last? Not forever. A thin market attracts competitors once the opening is visible. The agencies that keep their position will be the ones that keep publishing, keep earning third-party mentions and keep measuring.
Conclusion: the moat has moved
For years, large agencies could rely on size as a signal. Buyers assumed more people meant more capability, and Google rankings reinforced the idea. AI answers break that link in thin markets, because the model weighs the evidence it can find.
That is an opening for small agencies and a warning for anyone selling scale. The growth of one-person businesses, AI leverage for small teams and thin local answers all point the same way. The agency that publishes specific, sourced pages and measures the result can reach the shortlist.
To see where you and your competitors stand, use Airefs. It tracks which brands ChatGPT recommends for your prompts and shows the sources behind each answer. You can start with a 7-day trial at getairefs.com.