A growing number of people now ask an AI assistant for a recommendation before they ask a friend. Which estate planning attorney in a given city is well regarded. Which aesthetic practice specializes in a particular treatment. Which jeweler is known for a certain style of work. The assistant replies in a few sentences, often naming only a handful of options.
For the owner of a premium brand or professional firm, the obvious question is how those few names are chosen. The honest answer is that no assistant publishes a precise formula, and the systems differ from one another and change over time. But the broad mechanics are well understood, and they point to a clear set of priorities.
Two ways an assistant knows about you
It helps to separate two sources of knowledge, because they behave differently.
What the model learned in training
Large language models are trained on very large collections of text gathered from the web and other sources. From that text, a model forms a general impression of well-documented entities: what a company does, where it operates, what it is known for. This knowledge has a cutoff date and is not updated day to day. A brand that was widely and consistently described before that cutoff tends to be represented more fully. A brand that was obscure, inconsistently described or newly founded may be represented thinly or not at all.
What the assistant retrieves live
Many assistants can now search the web while composing an answer. Perplexity does this by design. ChatGPT, Gemini, Claude and Google’s AI Overviews also retrieve current information in many situations. When retrieval is involved, the assistant runs searches, reads a selection of pages, and synthesizes what it finds. This is where present-day search visibility matters directly: if your firm does not surface in the searches the assistant runs, it is unlikely to surface in the answer.
In practice, most answers draw on some combination of both. The training data provides a background impression, and retrieval provides current specifics.
The sources assistants tend to draw on
Whether through training or retrieval, assistants assemble their picture of a brand from the same broad categories of sources.
- Your own website. Your services, people, locations and the way you describe your work. Clear, well-structured pages are easier to interpret correctly.
- Business profiles and directories. Google Business Profile, professional directories, industry associations, licensing bodies and established listing sites. These often supply the basic facts: name, address, hours, specialties.
- Reviews. Ratings and, more importantly, the substance of what clients say. Recurring themes in reviews often find their way into how an assistant characterizes a business.
- Press and editorial coverage. Articles in respected publications, interviews, awards and features. Independent coverage helps establish that a brand is notable and credible.
- Reference sources. Encyclopedic sites, data aggregators and knowledge bases that many systems treat as relatively reliable.
- Discussion forums and community sites. Public conversations where people compare options and share experiences. These can carry real weight in some categories.
No single source decides the outcome. What matters is the overall pattern that emerges when they are read together.
Why consistency matters
When an assistant encounters conflicting information, it has to resolve the conflict somehow. It may choose one version, blend them, hedge, or leave the brand out in favor of one whose facts are clearer.
Common inconsistencies include:
- An old address still listed on directories after a move
- A firm name written several different ways
- Services described differently on the website, profiles and directories
- Former partners or practitioners still associated with the firm
- Closed locations that were never marked as closed
Each of these is minor on its own. Together, they make a brand harder to describe with confidence. Consistency is not a growth tactic in itself, but it removes friction that quietly works against you.
Why reputation matters
Assistants are generally designed to be helpful and to avoid steering people toward poor outcomes. When asked for a recommendation, they tend to favor options that appear well regarded across multiple independent sources.
For premium brands and professional firms, this has a few implications:
- Substance counts more than volume. Detailed, specific reviews that describe the actual experience tend to say more than a large number of brief ratings.
- Patterns surface. If many clients mention the same strength, that strength is likely to appear in how you are described. The same is true of recurring complaints.
- Independent voices carry more weight than your own. What a respected publication or professional body says about you is generally treated as more credible than what you say about yourself.
- Negative coverage persists. A prominent unfavorable article or unresolved dispute may be retrieved and summarized long after the fact.
Reputation in this sense is not something to be manufactured. Fabricated reviews and similar shortcuts violate platform policies, can carry legal risk, and tend to be detected. The durable approach is to earn genuine praise and make it easy for satisfied clients to share it.
A checklist you can act on
The following steps are within reach of most firms and brands, and they address the foundations that assistants rely on.
- Ask the assistants directly. Pose the questions your clients would ask to several assistants. Note whether you appear, how you are described, and which sources are cited. This is the starting point of AI visibility work, and of ChatGPT SEO in particular.
- Correct the basics at the source. Make sure your name, address, phone, hours and services are consistent across your website, Google Business Profile and the directories that matter in your field.
- State plainly what you do. Your website should say clearly, near the top of relevant pages, what you offer, who you serve and where. Avoid relying on taglines to carry meaning.
- Add structured data. Use schema markup to describe your organization, locations, people and services in machine-readable form.
- Keep your people current. Practitioner, partner and leadership pages should reflect who is actually at the firm, with credentials stated accurately.
- Invite genuine reviews. Make it simple for satisfied clients to leave honest feedback, and respond to reviews thoughtfully, including critical ones.
- Pursue relevant coverage. Seek earned mentions in publications, associations and directories that are genuinely respected in your field.
- Answer real questions in writing. Publish substantive content that addresses the questions clients ask before they engage you.
- Revisit regularly. Repeat the assistant checks every few months, since both the systems and the sources they draw on change.
Frequently asked questions
Why does an AI assistant recommend my competitor but not my firm?
Usually it is because the competitor is more clearly and consistently represented across the sources the assistant draws on, or because it surfaces more readily in live search. Reviewing which sources the assistant cites for your competitor is a useful place to start.
Can I ask an AI company to correct wrong information about my business?
Some platforms offer feedback mechanisms, but these are not a dependable fix. The more reliable path is to correct the inaccurate information where it originates, on your own site and on the third-party sources that assistants retrieve and learn from.
Do reviews really affect AI recommendations?
Reviews are one of the more visible signals of how a business is regarded, and their themes often appear in how assistants describe a business. They work alongside other factors, such as directory accuracy and independent coverage, rather than deciding the outcome on their own.