Most firms looking at ChatGPT advertising for law firms are about to make the same expensive mistake. They plan to copy a Google Ads account into a conversation, with the same keyword lists, bid caps, match types and negative keywords. Very little of that playbook carries over to an AI answer.
Google sells you a moment of intent. ChatGPT sells you a spot beside a conversation that is already half over by the time your ad shows up. These are two different products, and they reward two different skill sets.
The best PPC manager in town will not be the one who wins this channel. The winner will be the firm with the clearest point of view and the media to back it up.
Why Search Skills Break Down in ChatGPT Advertising for Law Firms
In search, the query is the unit of work. Someone types 'truck accident lawyer Dallas,' you bid on that string, and the auction decides who shows up. Every paid search skill grows out of that one fact: keyword research, match types, quality score, bid adjustments.
In ChatGPT, the unit is the whole thread. A prospective client does not type four words. They describe a situation over six or eight messages. They explain the crash, what the insurance adjuster said on the phone, the medical bills piling up, and how worried they are about missing work. By the time an ad appears, the person has already told the model more than they would tell your intake team in the first ten minutes of a call.
That changes the job. You are not matching a phrase. You are matching a stage of a problem. So map conversations instead of keywords. For a personal injury practice, a usable map looks like this:
- Shock: 'I was just in an accident, what do I do now?'
- Confusion: 'The insurance company called me, should I talk to them?'
- Evaluation: 'Do I need a lawyer for a soft tissue injury?'
- Comparison: 'How do contingency fees work and what is a fair percentage?'
- Decision: 'What should I ask a lawyer before I sign anything?'
Each stage needs its own message, its own offer and its own landing page. One ad that says 'Injured? Call Now' and runs across all five stages is a search ad dropped into the wrong room.
The Answer Is Not for Sale, So Your Ad Has to Agree With It
OpenAI has said publicly that ads do not shape what the model says. The answer is the answer. Your ad sits next to it, labeled as an ad, and gets judged against it in real time by someone who just read the model's explanation.
Picture the model telling a user to write down their injuries, avoid giving a recorded statement to the other driver's insurer, and talk to an attorney before signing a release. Right below that, your ad shouts 'Millions Recovered! Call 24/7!' You look like a billboard interrupting a doctor.
In a search result, your ad competes with other ads. In an AI answer, your ad competes with the answer itself, and the answer is usually more helpful.
Here is how to handle it: write the ad as the next logical step after a good answer. If the answer says to talk to an attorney before signing anything, your ad offers a free fifteen-minute review of the release before the person signs. If the answer explains contingency fees, your ad states your fee structure in plain numbers. The ad should read like the model would agree with it.
Your Landing Page Is Now Up Against a Better Explainer
Search visitors arrive with a question. Visitors from a chat arrive with an answer they have already read. They just got four hundred clear, patient, personalized words about their situation. Then they click through to a page with a stock photo of a gavel, three bullet points and a form asking for their case details.
They leave. Your page is not broken. It is just worse than what they read a few seconds earlier.
A landing page for conversational traffic has to go beyond the answer. It has to show what a chatbot cannot provide:
- A short video of the actual attorney explaining what happens in the first thirty days of a case like theirs
- A real timeline of your intake process, from first call to signed agreement
- The names and faces of the people who will pick up the phone
- A specific next step that fits the conversation stage they came from
People hire people. The piece on meeting the team behind the productions makes the point that the right people beat a big headcount. Your landing page should make the same case. A chatbot can explain the law, but it cannot introduce the paralegal who will return the client's call on a Tuesday night.
Creative That Works in a Chat Window Looks Nothing Like a Search Ad
Search ad copy is built for scanning: a headline packed with keywords, a description stuffed with benefits, and an extension for every phone number and location. That format exists because the user is skimming ten blue links and the ad has to grab the eye.
A chat user is not skimming. They are reading closely, in a calm interface, about a problem that matters to them. Keyword stuffing reads as noise there. The piece on why your ad is not underperforming and your creative is just boring argued that most weak campaigns have a creative problem, not a targeting problem. In a chat window that problem gets worse, because the creative is sitting next to the best-written text the user has seen all day.
Creative rules for this channel:
- Write in full sentences. The surrounding text is conversational, so your ad should be too.
- Lead with the specific situation, not the practice area. 'Offered a settlement within a week of your crash?' beats 'Experienced Car Accident Attorneys.'
- Make one offer per ad. A free release review, a fee explainer or a callback within an hour. Pick one.
- Cut the superlatives. 'Top-rated' and 'aggressive' are claims the user cannot check and the model did not make.
- Match the stage. Shock-stage ads offer calm and clarity. Decision-stage ads offer a direct path to a consultation.
Swapping headlines will not fix any of this. It takes a different writing brief, built around the conversation map, before anyone opens the ad platform.
Organic AI Visibility Decides Whether Anyone Believes Your Paid Placement
Here is the part search marketers miss entirely. When a user asks ChatGPT to name law firms that handle a certain type of case in a certain city, the model draws on what is published, cited and discussed across the web. If your firm is missing from that organic answer but shows up in the ad beside it, the user notices the gap. The ad looks bought because it is the only place your name appears.
Paid and organic visibility are separate line items in Google. In AI answers they work as one system. The organic answer tells the user whether your ad is worth believing.
Building that organic presence is a media job, not a bidding job. Models pull from material that exists in volume, shows depth and is attributed to a named expert. Here is what moves it:
- A podcast where your attorneys explain real legal questions in plain language, with full transcripts published on your own domain
- Video explainers filmed with the attorney on camera, each one tied to a specific question clients actually ask
- Bylined articles that take a clear position instead of restating the statute
- Guest appearances on other shows and publications, which put your name in front of new audiences and in sources the model can read
- Consistent firm details everywhere your name appears, so nobody confuses you with the firm down the street
This is why ChatGPT advertising and AI search visibility have to be planned together. A firm that buys placement without building authority is paying to send people back to an answer that does not mention it.
Measure ChatGPT Advertising for Law Firms by Signed Cases, Not Clicks
Conversational traffic does not act like search traffic. A user reads the answer, sees your ad, maybe clicks, then closes the tab and talks it over with a spouse. Two days later they search your firm by name on their phone and call. Last-click attribution credits branded search and calls the chat ad a failure.
If you judge this channel by cost per click or same-session conversion rate, you will shut it off before it ever has a chance to work. Set up measurement before you spend the first dollar:
- Use a dedicated tracking phone number on every ChatGPT landing page, separate from your search and TV numbers.
- Build dedicated landing pages for each conversation stage, so you can see which stage brings in clients and which brings in browsers.
- Ask one required intake question: 'Where did you first hear about us?' Record the answer exactly as the caller says it.
- Watch branded search volume and direct traffic for lift during flights, compared with the weeks before launch.
- Report on signed cases and case value over a ninety-day window, not leads in a seven-day window.
The firm that measures this channel like a TV buy, with its slow build and its effect on other channels, will understand what it is paying for. The firm that measures it like paid search will be wrong about it.
Bar Rules Follow You Into the Chat
A new channel does not mean new ethics. Every state bar's advertising rules apply to an ad in ChatGPT exactly as they apply to a billboard, a radio spot or a landing page. Misleading claims, guarantees of results, unverifiable superlatives and missing disclaimers are violations no matter where they show up.
The risk here is speed. Chat ad platforms make it easy to spin up dozens of variations quickly, and AI writing tools make it easier still. A firm that would never let a TV spot air without compliance review will happily push forty untested ad variations live on a Friday afternoon.
Run every chat ad variation through the same compliance review as your broadcast creative. Keep a library of approved claims, approved offers and required disclaimers by state, and write from that library. Speed is useless if a single variation lands you in front of the bar.
What Changes on Monday
ChatGPT is not a search box with a chat skin. It is a new media environment where the user is better informed, more patient and more skeptical than any search visitor you have ever paid for. Firms that treat it like Google will buy clicks from people who leave right away. Firms that treat it like media will build something their competitors cannot copy by raising a bid.
Here is the work for this week:
- Pull your last fifty intake calls and write down the actual questions clients asked before they hired you. That is your conversation map.
- Group those questions into five stages, from shock to decision, and write one offer for each stage.
- Audit your current landing pages against a good AI answer to the same question. If your page says less than the answer, rebuild it with the attorney on camera.
- Ask ChatGPT to name firms in your practice area and city. If you are not in the organic answer, start building authority media before you spend on placement.
- Set up tracking numbers, stage-specific pages and the intake attribution question before any ad goes live.
- Put every ad variation through compliance review, without exception.
The firms that move now will learn this channel while it is still uncrowded and while most competitors are still pasting keyword lists into it. That head start lasts only as long as the work behind it is real.