For most of the past two decades, small business owners had one question to worry about when it came to being found online: where do we rank in Google? That question still matters, but it is no longer the only one. A growing share of buyers now begin their research by asking an AI assistant. They open ChatGPT, Gemini, Copilot, or Perplexity, or they read the AI Overview at the top of a Google results page and ask for a recommendation in plain language. “Which CRM works best for a five-person sales team?” “Who repairs commercial HVAC systems in Portland?” “What is a reliable bookkeeping service for a small construction company?”
The answer they receive is not a page of ten links. It is a concise, confident shortlist, often three to five names, with a sentence or two explaining why each one was chosen. Businesses on that shortlist receive a warm introduction at the exact moment a buyer is ready to act. Businesses that are absent are not simply on page two. They are simply not part of the conversation, and they usually have no idea it happened.
Why AI recommendations behave differently from search results
Search engines return documents and leave the comparison to the reader. AI assistants often return something closer to a decision. They compress the field into a handful of options and present those options as the answer. Three characteristics of this behavior matter for a small business owner.
First, the shortlist is unstable. The same question, phrased slightly differently, can produce a different set of names. The same question asked in ChatGPT and in Gemini can produce two lists that barely overlap. Second, the assistant may not be reading your website at the moment it answers. It is drawing on how your business is described across the wider web: review platforms, industry directories, comparison articles, local news, forum discussions, and your own pages as they were last seen. Third, much of this activity may not appear in your analytics. A buyer who was told to call your competitor never visited your site, so there is nothing to measure in Google Analytics or Search Console.
The practical consequence is that a business can have a solid search presence and still be invisible in AI answers, or the reverse. The two channels overlap, but they are not the same channel.
What determines which businesses get named
When AI assistants recommend a business, they are usually reflecting a consistent, current picture assembled from multiple trusted sources. The businesses that appear tend to share a few traits. Their category is unambiguous: a reader, human or machine, can tell in one sentence what the company does, whom it serves, and where. Their facts are consistent everywhere: the same name, services, service area, and pricing model on the website, the Google Business Profile, LinkedIn, and industry directories. They have third-party validation: reviews, comparison articles, “best of” roundups, local press, and professional associations. And the material about them is recent enough to look alive.
Businesses that are missing usually fail on one of those points. The website describes benefits but never states the category plainly. Directory listings are outdated or contradictory. There is little or no third-party coverage, so the assistant has nothing independent to draw on. Or the coverage that exists is several years old and describes a product that has since changed.
A thirty-minute audit any owner can run
You do not need a specialist to find out where you stand. Set aside thirty minutes and work through the following steps.
Start by writing down ten questions a good-fit customer would realistically ask before choosing a business like yours. Include a mix: category questions (“best payroll service for restaurants”), comparison questions (“X versus Y for a small law office”), local questions (“commercial cleaning company near downtown Denver”), and problem questions (“how do I stop losing leads that come in after hours”).
Next, ask each question using at least three assistants, such as ChatGPT, Gemini, and Perplexity, and read the AI Overview in Google where one appears. For each answer, record which businesses were named, in what order, whether your business description was accurate, and which sources the assistant cited when it showed citations. A simple spreadsheet is enough.
Then look for the pattern. In most audits, the pattern becomes obvious quickly. Competitors that appear consistently are usually the ones with clear category language and steady third-party coverage. If you want a more detailed walkthrough of this process, this guide on how to check if ChatGPT recommends your brand covers the question set, the scoring, and the common mistakes in more depth.
Fixes, in order of impact
Once you can see the gap, close it in the following order.
Make your category impossible to miss. The first screen of your homepage and your About page should state, in plain words, what you do, who you serve, and where. “Outsourced bookkeeping for construction and trade contractors in the Pacific Northwest” gives an assistant something to work with. “We help you grow” does not.
Reconcile your facts. Audit your Google Business Profile, LinkedIn page, Yelp or industry-specific directories, and any association listings. Names, addresses, service lists, hours, and pricing language should match. Inconsistency reads as uncertainty, and assistants tend to skip uncertain options.
Earn independent coverage. Ask satisfied customers for reviews that describe the specific problem you solved, in their own words. Pursue inclusion in the comparison articles and roundups your buyers actually read. Where a trade publication or local outlet covers your industry, give them a reason to mention you: a data point, a case study, an expert comment.
Publish specifics. Pages that answer concrete questions with concrete details, such as pricing ranges, turnaround times, service boundaries, and worked examples, are more likely to be quoted by assistants. Vague marketing copy is not.
Re-measure. Repeat the ten-question check on a schedule and log the results. Movement in AI answers tends to lag your changes by weeks, so a single check tells you very little. A monthly log tells you a great deal.
Turning it into a routine
The owners who handle this well treat it the way they treat bookkeeping: a small, recurring task on the calendar rather than a one-time project. Thirty minutes a month, the same ten questions, the same spreadsheet, and a short list of fixes prioritized by what the answers reveal. Teams that want the measurement handled automatically can use a monitoring platform such as Seeno, which runs a brand’s real customer questions across the major assistants on a schedule, records which companies are named and how they are described, and shows where competitors are being recommended instead. Whether you use a spreadsheet or software, the discipline is the same: measure, fix, measure again.
The buyers have already changed how they ask. The businesses that will be recommended over the next few years are the ones that noticed early, checked where they stood, and did the unglamorous work of making themselves clear, consistent, and validated by independent sources. It is a manageable job for a small business, and it starts with ten questions and half an hour.