The short answer
For most family law firms, fix intake and tracking first, then earn AI citations (GEO) before paying for ChatGPT ads. OpenAI's ad policy (updated September 10, 2026) keeps ads out of sensitive user contexts, which it defines as personal, high-stakes, or emotionally vulnerable situations. Many divorce and custody conversations fit that description, so the organic answer may be the only place a firm can appear in them. A small, documented ChatGPT ads test makes sense once the firm is already named in AI answers, answers inquiries quickly, and is licensed in every state the ad would show in.
The short answer for family law firms
Start with the part your firm controls: answering inquiries and knowing where they came from. Then work on being named in useful AI answers. Buying an ad before those basics are ready adds a bill without telling you what worked.
GEO means generative engine optimization. It is work that helps AI tools understand and cite your firm, not a paid placement. Read what GEO means and our guide to earning AI citations for family law firms.
A free Visibility Report is a starting point. If your firm already appears in relevant answers and has working intake, a capped ads test can help you learn whether this separate channel fits. It cannot promise leads or recommendations.
Rented vs earned: what each one actually is
ChatGPT ads: a labeled unit below the answer
According to the OpenAI Help Center, "Ads in ChatGPT", ads appear below the answer and are clearly labeled. Free and Go users may see them. Plus, Pro, Business, Enterprise, and Edu have no ads. Ads are auction-priced, and paid delivery stops when spend stops.
GEO: being named or cited inside the answer
A citation is part of the answer, not the ad below it. You cannot buy that mention. OpenAI says ads "do not influence ChatGPT's answers" and run on separate systems. Clear firm information, useful answer pages, and accessible sources support this work, but no agency can guarantee a citation.
Compare the two surfaces
| What you get | Who sees it | When it starts/stops | How it's measured | Family-law caveat |
|---|---|---|---|---|
| ChatGPT ads: a labeled unit below the answer | Free and Go users in eligible contexts | After approval and auction delivery; stops when spend stops | Clicks, consultations, and signed cases | Sensitive conversations may not show ads |
| GEO: a firm named or cited inside the answer | People reading an AI answer, not only ad-supported tiers | No fixed start date; mentions can change or disappear | Repeat prompt checks and intake source reports | A mention cannot be bought or guaranteed |
Can family law firms advertise on ChatGPT?
Yes, in the US if the advertiser is licensed where the ad is shown, with manual, case-by-case approval. OpenAI's Ad policies, updated September 10, 2026, say:
"Ads for legal advice, representation, or legal services offered to individuals or businesses are permitted in the US only when the advertiser is licensed to practice law in the jurisdiction where the ad is shown. This includes services related to immigration, personal injury, legal claims, or document preparation."
Legal-services ads outside the US are currently prohibited. Regulated categories are rolling out gradually with manual approvals, so permission in the policy is not the same as an approved account or ad.
The changelog says legal services became permitted in the US in August 2026 (policy v1.5). Older blanket claims that "legal ads are banned" do not reflect that update.
The live page still has mixed launch-era wording. One section lists financial or legal services among categories "disallowed at launch," while the Legal services section and v1.5 say US legal services are permitted. That is another reason to re-check the live policy and get approval before spending.
The family law problem: sensitive conversations
What OpenAI's policy says
The policy's sensitive-context safeguard says: "Our safeguards are designed to prevent ad placements in sensitive user contexts." It defines those contexts as "conversations involving personal, high-stakes, or emotionally vulnerable situations where ads could undermine user trust or create a poor experience."
The Help Center also asks, "Will ads appear in sensitive chats?" Its answer is: "No. Ads are not eligible to appear near sensitive topics, including personal health, mental health, or politics."
The policy does not name divorce, custody, or family law. This is not a claim that OpenAI bans ads in divorce chats. It is a reason to expect that some valuable family law conversations may not be eligible, and to test rather than assume.
Why divorce, custody, and protective-order chats may fit
Many such conversations are personal, high-stakes, and emotionally vulnerable, the policy's own definition. A person asking about safety, their children, or leaving a marriage may be deciding whether to contact a lawyer right then. A general research question and a detailed personal crisis are not the same setting for an ad.
What it means for the budget
Carlos Arias, September 23, 2026, frames it this way: "So paid buys you research mode. Organic covers research mode and the crisis." That is his interpretation, not a promise of delivery from OpenAI.
Ads may reach research-mode conversations. Where ads are withheld, the organic answer remains the possible place for a firm to be named. That does not guarantee the answer will name any firm, including yours.
How to check without assuming
Run your own prompt set on a Free account on different days. Include general research questions and realistic, fictional scenarios without sharing client details. Note whether an ad appears and whether your firm is named or cited. Keep the date and prompt so you can repeat the check. One ad-free answer does not prove a permanent rule.
What one law firm actually paid (and what it doesn't prove)
WKND's October 23, 2026 report describes a San Francisco law firm test from June 15 to July 31, 2026 (47 days). The firm spent $11,982 on ChatGPT Ads and received 114 leads, at $105.11 per lead. Its own Google Search cost per lead was $248.
WKND reports ChatGPT cost per click at about $2.62, with a 2.49% click-to-lead rate versus 17.43% for Google. Leads per $1,000 were 9.5 for ChatGPT and 4.0 for Google.
On those costs alone, ChatGPT leads would need to sign at about 42% of Google's signing rate to match cost per signed case. That is break-even math, not a published result showing those leads signed.
This was one firm over 47 days. The page does not state its practice area or publish signed-case figures. The $248 Google figure belonged to that firm. Do not treat any of these numbers as a family law benchmark, a family law cost per lead, or an Up And Social result.
WKND presents ChatGPT Ads as a third surface, not a replacement for Google or organic visibility. A lower lead cost is useful only if the inquiries fit the practice and become clients at an acceptable cost.
The state-bar layer
OpenAI approval does not replace your state's attorney advertising rules. Licensing is also a targeting question: confirm Ads Manager can restrict delivery to the states where the advertiser is licensed before you run a test.
- Filing states: review requirements before testing new copy. Filing and review can slow creative changes.
- Disclosures: confirm required information fits and remains readable in the small ad unit.
- Records: keep copies of ads, dates, approvals, targeting settings, and changes.
- Solicitation: a contextual ad under a user's detailed account raises an open question under Model Rule 7.3. Do not assume ordinary advertising rules settle every placement.
Ask ethics counsel whether the proposed placement is advertising or solicitation under your state's rules. Do not treat this article as a bar ruling. See attorney advertising rules for divorce ads for Florida, Texas, and New York specifics.
Which to fund first: a decision table for family law firms
Choose the next step from your firm's current position, not from a new platform's headline. These are starting points for a measured decision, not promises of results.
| Situation | Fund first |
|---|---|
| Calls go to voicemail after hours / no source question at intake | Fix intake and tracking first |
| Not named by ChatGPT for core questions | GEO first |
| Already named, Google CPL is high, licensed in target state | Capped ChatGPT ads test beside Google, same dates |
| Small budget | Low-cost GEO (entity cleanup, answer pages, crawler check) first |
| Multi-state firm | Confirm state-level targeting in Ads Manager before any test |
| Estate planning/probate firm | Research-mode conversations may suit ads better than crisis divorce; still test |
For a small GEO budget, start with accurate firm and attorney information, clear answer pages, and a check that crawlers can reach them. For an estate planning or probate firm, research-mode questions may be a better fit for ads than crisis divorce questions. That is a reason to test, not a proven comparison.
How to measure each side
Ads: follow the spend to consultations and signed cases
Use UTM tags and conversion tracking to connect ad activity with consultations and signed cases. Compare the same dates and case types where possible. Cost per click and cost per lead are intermediate numbers. Our Google Ads cost-per-signed-case math explains why the final step matters.
GEO: check mentions and ask where inquiries came from
Repeat a monthly prompt set in ChatGPT, Google AI, and Perplexity. Note whether your firm is named or cited and which page is cited. Answers change, so a single screenshot is not lasting proof.
Add "How did you find us?" at intake, with ChatGPT, Google AI, and Perplexity as separate answers. This is a measurement recommendation, not a claim about fields Up And Social has already installed. Self-reported sources have limits, but they help connect mentions with real inquiries.
Use a Visibility Report as a baseline. Our family law lead generation guide covers intake speed and missed inquiries.
Where Up And Social fits
We help family law firms with both paid search and AI search visibility, and we report them separately. We have not yet published client results for ChatGPT ads.
No honest agency can guarantee a ChatGPT recommendation. Our family law AI search page explains what we can work on and what we cannot control.
Our published programs on the pricing page are Activate $1,750, Elevate $2,750, Dominate $4,750 per month, plus a one-time $1,497 activation fee. Paid media is separate. These are our published programs, not a new ChatGPT ads package or price.
For the Google budget timing side, read when to start family law Google Ads. Treat Google and a possible ChatGPT test as separate decisions with separate records.
Frequently asked questions
Should a family law firm fund ChatGPT ads or GEO first?
For most family law firms, fix intake and tracking first, then earn AI citations (GEO) before paying for ChatGPT ads. OpenAI's ad policy (updated September 10, 2026) keeps ads out of sensitive user contexts, which it defines as personal, high-stakes, or emotionally vulnerable situations. Many divorce and custody conversations fit that description, so the organic answer may be the only place a firm can appear in them. A small, documented ChatGPT ads test makes sense once the firm is already named in AI answers, answers inquiries quickly, and is licensed in every state the ad would show in.
Can family law firms advertise on ChatGPT now?
Yes, in the US if the advertiser is licensed where the ad is shown, subject to manual, case-by-case approval. OpenAI's August 2026 policy v1.5 says US legal services are permitted. Only Free and Go users see ads. Re-check the live policy and account eligibility before spending.
Will a ChatGPT ad show when someone asks about their divorce or custody case?
You cannot know in advance. OpenAI does not name divorce or custody in its policy, but many such chats fit its definition of sensitive contexts: personal, high-stakes, or emotionally vulnerable. Test a prompt set on a Free account on different days rather than assuming an ad will appear.
Do ChatGPT ads change what ChatGPT recommends?
No. OpenAI says ads do not influence ChatGPT's answers and run on separate systems. Paying for an ad does not buy a recommendation or citation inside the answer.
Who actually sees ChatGPT ads?
Free and Go users may see ads. Plus, Pro, Business, Enterprise, and Edu users do not. Ads are not shown to users under 18 or in Temporary Chats, and sensitive conversations are not eligible.
What has a law firm paid per lead?
WKND reported $105.11 per ChatGPT Ads lead versus that firm's $248 Google Search cost per lead in a 47-day San Francisco law firm test. The practice area was not stated and signed-case figures were not published. This is not a family law benchmark or an Up And Social result.
How do you measure GEO with no CPC?
Check the same prompt set each month in ChatGPT, Google AI, and Perplexity. Record whether your firm is named or cited. Ask how people found you at intake, with those tools as separate answers. A Visibility Report can give you a starting baseline, not a guarantee of future mentions.
What state-bar issues come with ChatGPT ads?
Review licensing and state-level targeting, filing requirements, disclosures, and record-keeping. The line between contextual advertising and solicitation raises an open question under Model Rule 7.3. Ask ethics counsel to review the proposed placement and copy before a test.
What should a firm do first before either?
Fix intake and tracking. Confirm who answers inquiries, how missed calls are handled, and how you record the source, consultation, and signed case. More exposure is not useful if you cannot answer or measure the inquiry.
Related guides
- GEO vs AEO vs AIO
- AI search for family law firms
- AI marketing for family lawyers
- AI Overviews for family law firms
- Divorce lawyer advertising
- When to start family law Google Ads
- AI search optimization for law firms
This article is general information for law firm owners. It is not legal or ethics advice. Check your state's attorney advertising rules before running any ad. Prior results do not guarantee a similar outcome. No agency can guarantee rankings, leads, or AI recommendations.