Paid Directory Listings Can Turn Into Organic ChatGPT Recommendations

I recently asked ChatGPT to recommend a car accident lawyer. I do this periodically on the free version of ChatGPT to see how our clients are being recommended to ChatGPT’s general user base. It gave me five firms with star ratings, review counts and a short description of each. The answer was clean and confident, and it read like advice.

Then I asked where the list came from. It turns out, paid placements on legal directories can become organic ChatGPT recommendations, and the person reading the answer has no way of knowing.

What the Answer Was Built On

The source chips under the answer showed Avvo plus one other site. When I pressed, ChatGPT named the second source: Expertise.com. We already know that Avvo is one of the directories that ChatGPT cites, as we covered that last year.

A few questions later, ChatGPT agreed that its recommendation was much closer to repeating prominent entries from third-party lists than to independent research. In its words, it had taken search prominence and directory information and turned it into language that sounded like its own judgment.

Then its explanation changed. I opened the Avvo page myself, and the firm at the top of ChatGPT’s list wasn’t on it. The page was sorted by “Best Match,” and the top listings carried PRO badges. ChatGPT backed off its earlier explanation and said it couldn’t reconstruct what it had seen. The top firm eventually turned up on Expertise, but on that directory’s pages for neighboring practice areas in the same city, not the car accident page.

So the model couldn’t reliably explain its own recommendation. But it could identify the kind of sources it had used, and that’s where this gets interesting for law firms.

What the Directories Say About Their Own Rankings

Neither directory hides its business model. Expertise’s terms of use describe the site as supported by advertising and affiliate links, and they state that compensation may affect how and where providers appear, including their order on the page. The same terms say Expertise may sell “Featured” and “Sponsored” ads that include preferred placement, and they also say most of the experts it lists don’t pay anything. The city page ChatGPT drew from included its own disclaimer about featured lawyers paying an advertising cost, along with a “Featured Provider” label.

We also regularly encourage our law firms to get listed on Expertise when they’re given the opportunity.

Avvo is just as direct. Its advertising page for lawyers says sponsored listings put an attorney at the top of directory pages when consumers search for a lawyer.

None of this is improper. Bar rules generally let lawyers pay the reasonable costs of advertising, and a directory listing is advertising. On the directory itself, the “Featured,” “Sponsored” or “PRO” label sits right next to the name, and the reader can weigh it.

The problem is what happens after ChatGPT reads the page.

The Label Doesn’t Make the Trip

When ChatGPT pulls from a directory and rewrites the results as “I’d start with these firms,” the label disappears. A listing that was marked as advertising on the source page becomes a recommendation in the answer. The consumer sees a list that looks like the AI’s own assessment, with nothing to show that the underlying ranking may have been shaped by who paid to be on the page that the AI model just read.

OpenAI’s current ad policy permits ads for legal services in the US when the advertiser is licensed to practice law in the jurisdiction where the ad appears. Those ads are labeled and separated from the answer. But answer independence covers OpenAI’s ads. It says nothing about the commercial arrangements built into the web pages ChatGPT reads. That leaves a law firm with two ways to put money in front of a ChatGPT user. One is a labeled ad at the bottom of the response. The other is directory placement that may surface, unlabeled, inside the answer itself. That’s what happened in this experiment.

What This Means for Your Firm

Your directory spend has a second audience. Directory pages used to be read only by people. Now they’re also read, summarized and repeated by answer engines. When you evaluate a directory budget, think about both audiences.

Find out which sources AI cites in your market. Run the questions your clients would ask. Do it logged out, more than once, and write down the sources that show up under each answer. If the same two or three directories keep appearing, your presence on them is part of your AI visibility whether you pay or not. We track this for clients as part of our answer engine optimization work.

Don’t mistake a directory buy for an AI strategy. You don’t control whether ChatGPT reads the page, which page it reads, or how it describes you. In my test, the top firm came from a directory page for a different practice area. Paying for placement might put you in the answer. It might not. You won’t know which, and you won’t know why. Also, with each new model, the rules change.

Your own website still does the heavy lifting. When I asked ChatGPT to rebuild the list without the directories, it went to the firms’ own websites. It looked for founding history, published case results, attorney bios and practice pages. A firm whose site clearly documents its experience gives the model something to cite that isn’t someone else’s paid list. Basically, the first time around, it took a shortcut. The second time, it actually did a little work.

Keep every listing accurate. In the same conversation, one firm’s satellite office profile showed a handful of reviews and a low average, which contradicted a much larger review record elsewhere. AI tools grab whatever structured listing they find. A neglected profile can speak for your whole firm.