We Asked ChatGPT to Recommend Local Businesses in Portland: What 60 Answers Show

We asked an OpenAI model with live web search 60 questions a Portland customer might type, from "best plumber in Portland, Oregon" to "best hair salon in Gresham with good reviews", and logged every business it recommended. The answers lean heavily on Google-style business data: 93% of them showed star ratings or review counts, the businesses named were almost all different from one phrasing to the next, and the first name on the list stayed the same in only 12 of 20 repeated questions. This is the data behind our guide to Portland SEO.

Three things to know before you go further:

• Ratings are front and center. Of 249 recommended listings that showed a rating, 214 (86%) scored 4.5 or higher and the average was 4.71.
• You do not need a huge review count. The median recommended business had 69 reviews, and 104 of the 249 had fewer than 50.
• The answer is not stable. Ask the same city question twice and the top name changed in 8 of 20 categories.

How we ran the study

We built 20 service categories across home services, health, legal, finance and personal care: plumber, electrician, HVAC company, roofing contractor, house cleaning service, moving company, landscaper, painter, pest control company, auto repair shop, dentist, chiropractor, veterinarian, therapist, personal injury lawyer, immigration lawyer, accountant, real estate agent, mortgage broker and hair salon. Each category was asked three ways: a city question ("Who is the best plumber in Portland, Oregon?"), a neighborhood question (Pearl District, Sellwood, Alberta Arts, Hawthorne, St. Johns) and a suburb question (Beaverton, Hillsboro, Lake Oswego, Tigard, Gresham). That is 60 prompts, all answered on October 6, 2026.

Method and limits, so you can judge the numbers fairly: we used the OpenAI API (model gpt-4.1-mini) with its built-in web search tool set to a Portland, Oregon location, not the consumer ChatGPT app. The app can behave differently by account, memory and model version. A second model pass listed the businesses named in each answer, in order. It is a single run on one day, so treat it as a snapshot, not a ranking of any business. We do not name individual businesses in this article.

What the answers look like

The first pattern is breadth. A city question named 5.8 businesses on average, a suburb question 4.8 and a neighborhood question 4.2. The more specific the place, the shorter the list, which means the competition inside a tight neighborhood is a handful of businesses, not dozens.

Ratings and reviews show up in almost every answer

Almost every answer came with the same kind of data you see in a Google Business Profile: a star rating, a review count, an address and an open or closed status. In the 249 listings where a rating appeared, 240 (96%) were 4.0 or higher and 214 (86%) were 4.5 or higher. The average was 4.71.

Review volume was less of a barrier than we expected. The median was 69 reviews, 108 listings had 100 or more, and 104 had fewer than 50. A newer business with a strong rating and a modest review count still gets named, as long as the profile is complete and the rating is high. For the full local ranking picture, see our Portland SEO guide.

See if ChatGPT recommends your Portland business Analyze your AI Search Score without login - no account needed. Start getting recommended by ChatGPT and Google for your Portland business - Start for Free - 3 articles published in 3 days. Start for Free

Where the answers get their information

Every answer cited at least one source, and in our run the cited pages were the businesses' own websites far more often than review directories: none of the 60 answers cited Yelp, Angi, Thumbtack or the Better Business Bureau as a link, and none cited Reddit or a local news site. Google Maps links appeared in 5 answers. That does not tell us where the rating data comes from, but the link a customer is invited to click is usually the business's own site. A clear, crawlable website with the services and the city stated in plain words is the page the model points to.

The answer changes from run to run

We repeated the 20 city questions in a second pass. The first business named was the same in 12 of 20 (60%), and on average only 56% of the businesses named in the first run appeared again in the second. Within a category, the three phrasings produced between 10 and 17 distinct businesses, and no single business appeared in all three.

The practical meaning: you cannot "hold" a position in ChatGPT the way you can track a rank in Google. What you can do is raise the odds of being in the set. That means a complete Google Business Profile, a steady flow of reviews and a website that answers the question in the words customers use. Pages that open with a direct answer and state the city and the service are easier for a model to lift into a recommendation.

What to do with this

Our study shows what the answers display, not why a business was chosen, so these are sensible actions that match those signals:

• Complete your Google Business Profile. Categories, hours, address, services and photos feed the exact fields the answers display.
• Keep your rating above 4.5. 86% of rated recommendations cleared that bar. Ask every happy customer for a review.
• Write a page per service and per area. Neighborhood and suburb questions return different businesses than the city question.
• Publish helpful answers on your own site. The cited link was usually the business site, so the content on it is what gets read.
• Measure it, do not guess. Because answers vary between runs, track how often you appear across many questions, not once.

If you want the pricing, timeline and ranking side of this, start with our full Portland SEO guide. To see how visible your own business is to ChatGPT right now, use the free AI Search Score below.

Frequently Asked Questions: ChatGPT and local businesses

Sources

Focus on these:

• Rocketito study data: 60 Portland local prompts and a 20 prompt repeat, run on October 6, 2026 with the OpenAI Responses API and web search tool
• BrightLocal: Local Consumer Review Survey