Logo
Check Lost Sales

How chatgpt decides what brands to recommend | yazeo

ChatGPT doesn't recommend randomly. It evaluates specific signals to decide which businesses to name. Learn what those signals are and how to build them.

Introduction

Every time someone asks ChatGPT "What's the best [anything]?", ChatGPT makes a decision. It evaluates what it knows, searches the web if needed, and picks specific businesses to name. Your competitor gets chosen. Or you do. There's no middle ground.

This isn't random. ChatGPT follows identifiable patterns when deciding who to recommend. Understanding those patterns is the difference between hoping AI notices you and building the evidence that makes AI choose you.

After analyzing thousands of ChatGPT recommendations across industries, the Yazeo team has mapped exactly how ChatGPT evaluates businesses. This guide shares what we've found.

Chatgpt uses two systems to form opinions about your business. most businesses only influence one.

System 1: Training data. ChatGPT's base knowledge comes from massive amounts of internet text it was trained on. If your business had a strong, widespread digital presence during the training period, ChatGPT has a favorable baseline understanding of you. If your business was barely mentioned anywhere online, ChatGPT essentially doesn't know you exist.

Training data updates happen when new model versions are released. These are infrequent (a few times per year) but produce significant shifts. Businesses that built digital authority between training updates see sudden improvements when the new model launches.

System 2: Real-time web search. For many queries, especially those requiring current information, ChatGPT searches the web live and incorporates what it finds into its response. This means what's on the web right now influences what ChatGPT says right now.

This is critical because it means you can influence ChatGPT's recommendations without waiting for the next model update. A new review, a fresh article, an updated directory listing, recently published content. All of these can shift what ChatGPT says about your business today.

Most businesses only think about System 1 ("we need to be in the training data"). The Yazeo ARO System targets both, building long-term authority for training data and current web presence for real-time search.

Want to see exactly what ChatGPT says about your business right now?

See who ChatGPT recommends instead of you. Free. Instant.

Check AI Competitors

See who ChatGPT recommends instead of you. Free. Instant.

When chatgpt picks a business to recommend, this is what it evaluates.

When someone asks "Who's the best personal injury lawyer in Houston?" ChatGPT doesn't have an internal ranking list. It evaluates available evidence in the moment.

What credible sources say about you. If multiple trusted publications, directories, and review platforms mention your business favorably, ChatGPT registers that as independent validation. A business mentioned on three industry publications carries more weight than one mentioned on none, regardless of how good their own website copy is.

How consistent your information is. If ChatGPT finds conflicting information about your business (different descriptions, different addresses, different service lists on different platforms), it can't form a confident recommendation. It defaults to a competitor whose information is cleaner.

Whether your content answers the specific question asked. If someone asks about "personal injury lawyer for truck accidents," ChatGPT looks for businesses with content specifically addressing truck accident cases, not just general personal injury content. Specificity wins.

Your review profile across platforms. ChatGPT evaluates reviews not just for star ratings but for recency, detail, volume, and platform distribution. Recent, detailed reviews on multiple platforms signal a healthy, active business.

Whether it can understand your business clearly. Structured data on your website tells ChatGPT exactly what your business is in machine-readable format. Without it, ChatGPT interprets your website content and often misinterprets or ignores it.

The difference between being mentioned and being recommended.

There's an important distinction in how ChatGPT handles businesses. Being mentioned ("Some firms in Houston include X, Y, and Z") is different from being recommended ("For truck accident cases in Houston, I'd recommend X because...").

Mentions happen when ChatGPT has some awareness of your business but not enough confidence to endorse you. Recommendations happen when ChatGPT has enough evidence to feel confident putting its credibility behind your name.

The gap between mention and recommendation is confidence. And confidence comes from the five ARO signals working together: content depth, review strength, data consistency, third-party authority, and technical structure.

In most industries, 2 to 3 businesses capture over 70% of AI recommendations for their category. Those businesses didn't just get mentioned. They built enough evidence that ChatGPT feels confident actively endorsing them. That's the position the Yazeo ARO System builds.

Chatgpt's recommendations aren't permanent. they shift based on new evidence.

The brand ChatGPT recommended last month might not be the one it recommends today. Several things trigger changes.

New reviews on trusted platforms. A surge of positive, detailed reviews can push a business into recommendations it was previously absent from. A cluster of negative reviews can drop a business out.

New third-party coverage. Getting featured on an industry publication, a "best of" list, or a major review site can shift ChatGPT's real-time recommendations within days.

Competitor activity. If a competitor publishes strong comparison content, earns a review milestone, or gets featured on a trusted source, they can displace your business from recommendations you previously held.

Content freshness. ChatGPT favors current information. Content published three years ago that hasn't been updated gradually loses influence compared to recently published material on the same topic.

Model updates. When OpenAI releases new model versions with updated training data, recommendation patterns can shift broadly.

This dynamic nature is why AI Recommendation Optimization (ARO) is an ongoing process, not a one-time project. The Yazeo ARO System monitors every shift and responds before competitors take advantage.

How chatgpt's decision-making plays out in real business outcomes.

Immigration law firm, Miami FL. When we first checked, a firm with 18 years of experience appeared in zero ChatGPT recommendations. A three-year-old competitor appeared in 64% of immigration queries in the metro. The reason: the competitor had stronger third-party mentions, more recent reviews, and cleaner structured data. Not better credentials. Stronger signals. Within 120 days of the Yazeo ARO System building all five signals, the established firm appeared in 41% of tracked queries. AI became 22% of new consultations. Revenue: $89,000 in retained fees from AI-referred clients in one quarter.

ChatGPT didn't care which firm was "objectively better." It cared which firm it could verify more confidently.

Questions about how chatgpt makes recommendations.

ChatGPT is making a decision about your business every time someone asks.

Do you know what it's deciding? Find out instantly. See what ChatGPT says, who it recommends instead, and what's driving the decision.

Am I on ChatGPT? Find out.

Most popular pages

Industry AI Search

How Immigration Attorneys Can Get Recommended by AI Search Engines

He is a software engineer from India on an H-1B visa. His employer has started the green card process through the PERM labor certification, but he has questions he is not ready to ask a lawyer yet. He is not sure how long the process will take from his country of birth's priority date, whether he should file I-485 adjustment of status concurrently, and what happens to his H-1B status if his employer terminates his employment during the process. He opens ChatGPT and asks three specific questions about the EB-2 and EB-3 backlog for Indian nationals. ChatGPT explains the priority date system, the visa bulletin, the current cut-off dates for India, and the distinction between filing when a date becomes current versus filing concurrently. He asks a follow-up question about portability under AC21. Then he types: "Best immigration attorney near me in [city] who specializes in employment-based green cards for Indian nationals, H-1B extensions, AILA member." ChatGPT names two firms. He schedules a consultation with the first. Your firm has an attorney with 12 years of employment-based immigration experience, handles EB-2 and EB-3 cases for Indian nationals regularly, is an AILA member, and has a strong Google review profile. ChatGPT named someone else. Not because your attorney is less qualified. Because the two firms it named had documented their employment-based immigration specialization, Indian national experience, AILA membership, and visa-specific expertise in AI-readable formats, and yours had not.

Industry AI Search

How Dermatologists Can Show Up in AI Search Results and Get More Patients

<p>When a patient notices a suspicious mole, their first instinct is no longer to call their primary care doctor for a referral. Increasingly, they open ChatGPT and type: "Best dermatologist near me for skin cancer screening" or "Top-rated dermatologist in [city] for acne treatment." The AI gives them one to three names. If your practice is not one of them, that patient books with whoever the AI recommended and you never know they were looking.</p><p>Dermatology sits at the intersection of two categories that AI search is disrupting fastest: healthcare (where 49% of Google searches now trigger AI Overviews) and aesthetics (where high-consideration patients research extensively before booking). Your practice probably treats both medical and cosmetic dermatology patients, and both patient types are shifting toward AI-driven discovery.</p><p>The challenge for dermatologists is that AI platforms need to understand what kind of dermatology you practice, which conditions you treat, what procedures you offer, and where you are located, all from structured, consistent, extractable information spread across the web. A practice that says "We treat all skin conditions" gives the AI nothing specific to recommend. A practice that has dedicated pages answering "What is the best treatment for cystic acne in adults?" and "How often should you get a full-body skin check?" gives the AI exactly what it needs.</p>