Most agencies will hand you a keyword list for ChatGPT ads. I would start with the moment a person stops researching and starts picking a provider. This list is for lead-gen businesses and the people running their tests: plumbers, lawyers, clinics, home services, and anyone else who needs booked work rather than interesting traffic. Each of these nine conversations needs its own offer and landing page. Test two or three first, not all nine.
1. Show up when they ask who to call
This is the closest thing here to high-intent search. The prompt sounds like “who’s a good divorce lawyer in Austin” or “recommend an emergency plumber near me.” The person has stopped learning about the problem and started choosing someone to solve it. BrightLocal reports that 45% of local consumers use ChatGPT or AI tools for local business recommendations, up from 6% in 2025.
The planning unit is the conversation, not the keyword. Instead of building around plumber near me, you can provide a short context hint such as “homeowner with burst pipe needs plumber fast.” Context hints help match ads to a live chat semantically; questions in a dialogue reveal both the reason someone is looking and the stage they have reached. That gives you a more useful way to plan the offer. It does not mean you get to buy a place in ChatGPT’s recommendations.
Start with an ad that names the service and city plainly. Send the click to a page with three proof points, a service-area map, a phone number above the fold, and a two-field quote form. Offer a clear reason to call you now, not a blog post. The risk is comparison shopping; expect the buyer to check another option. On a small budget, I would build this page first.
2. Take the emergency call, if you can answer it
“Water coming through ceiling, who do I call now” is not a request for a guide to plumbing. Panic shortens the shortlist to whoever can respond. That makes the emergency conversation valuable, but it also punishes slow follow-up harder than almost any other entry on this list.
The offer is simple: a 60-minute response if you can deliver it, a live phone number, and an after-hours fee stated upfront. The landing page needs a click-to-call button, a service-area list, and three lines on what to do before the technician arrives. Route the click to a phone, not a form. Forms die in emergencies.
Lead quality is not the main risk here. Capacity is. An ad running at 2 a.m. is a bad investment if nobody answers at 2 a.m. This would be my second test after the provider shortlist, but only for a business that can take the call and do the work. Check that first; write the ad second.
3. Offer a real appointment when they ask who is available
“Dentist open Saturday near me taking new patients” and “garage door repair available tomorrow morning” have a different filter from “who’s best?” The buyer has already decided to book. They need to know who has time.
Your offer is a bookable slot, not a consultation. Show the next three openings on a mobile-friendly landing page, along with your service radius by ZIP code and a booking widget that works on a phone. If availability is the deciding factor, a page that says “call for availability” makes the buyer do the work your ad promised to remove.
Stale calendars are the quality risk. You may buy a good click and lose the job because the time on the page is no longer open. This is a strong third test only if your team can confirm an appointment in under 15 minutes. This home-services account’s lead-quality story makes the same practical point from Google Ads: structure and follow-up determine whether useful clicks become booked jobs. Before you pay for availability intent, make sure your availability is real.
4. Give price shoppers a way to get their number
I used to tell clients to avoid price clicks. I was wrong to treat every cost question as idle curiosity. Someone asking “how much does a full panel replacement cost in Dallas” has at least named a real, billable problem. The chat may give them a broad range; your ad can offer a way to narrow it.
Offer an estimator or photo quote, not “contact us.” The page should show real price ranges, explain what moves the price up or down, and ask for the three qualifiers you actually use to price the job. If a visit has a fee, say so. Hiding it to collect another form fill does not make the lead better.
Some visitors will still want free advice. Let the form and the stated visit price separate them from people ready to proceed. A buyer who has already faced the likely cost needs a next step, not another vague invitation to talk. Build the estimator before buying the cost-question click.
5. Catch the point when troubleshooting becomes a service call
A diagnosis chat can run for a while: “water heater pilot keeps going out, tried relighting twice.” At first, the person wants to fix it alone. The useful moment comes when the conversation turns from troubleshooting to whether they need a professional. An ad for general water-heater advice arrives too early; an ad for a diagnosis can meet the decision.
Offer a same-day diagnosis with a fee credited toward repair. The landing page should repeat the symptom in plain language, describe the first two fixes you will check, and show a calendar instead of a whitepaper. That is a page for someone weighing a service call, not someone collecting maintenance tips.
The risk is curiosity volume. Keep context hints narrow around “needs a pro” language, and let the page disqualify people who still want to DIY. Small budgets cannot afford to learn that every plumbing question is not a plumbing lead. The discipline in structuring and bidding for local-service Google Ads applies here too: one poor match can eat into a test before the right person sees it. Start with a narrow symptom and a page built for that symptom.
6. Give comparison shoppers a decision rule
“ABC Plumbing vs XYZ Rooter, which is better for trenchless sewer line” tells you what a generic service query does not: the buyer has narrowed the field and wants a tiebreak. You do not need to trash the other firm. You need to explain the tradeoff that matters to this buyer.
Offer a one-page comparison written by you, with a booking option on the page. Put the decision rule in the headline: response time, warranty, or price. Then put your proof under each relevant point. Only pay for comparisons you can win on the stated criteria. If the buyer wants the cheapest option and you are not it, skip that conversation.

The lead risk is paying to help someone choose the other firm. A specific comparison page gives you a better chance than sending that person to your homepage and hoping they work out your advantage. Pick one tradeoff you can substantiate, then build the page around it.
7. Meet unhappy customers at the switching point
“My cleaning company keeps missing spots, should I switch?” and “my agency isn’t reporting, what should I look for?” are not category-research prompts. The buyer already has a provider. They are deciding whether the inconvenience of replacing them is worth it.
What a deck might call churn intent, I call a complaint with a next step attached. Offer a switching audit with no restart fee. The landing page should name the complaint back: missed visits, no-shows, or vague invoices. Then show onboarding in three steps and explain how quickly you can take over. Make the move feel manageable without pretending it requires no work.
The risk is expectations. A switcher who sees the same problem twice will not thank you for a polished ad. This can fit a small budget for home services or agencies because the prompt is specific, but keep the context hint tight around dissatisfaction. Otherwise you will pay for people seeking advice on handling a difficult provider, not people ready to hire a new one. Write the onboarding steps before the ad copy.
8. Turn a compliance question into a check
“Do I need a permit to replace a water heater in Chicago?” and “is this cosmetic procedure covered if I have this diagnosis?” do not sound like classic ad queries. They do describe a person who needs to know what applies before paying for work. If guessing wrong could cost them money or create a legal problem, a clear explanation earns attention.
Offer a compliance check, not a generic quote. The page should state your license number where relevant, give the applicable rule in one sentence, explain what paperwork you handle, and ask for the address and timeline you need to assess the job. Check that rule before you publish the page; the ad is not a substitute for getting it right.
Some people only want the free answer. A paid site visit credited toward the job can separate them from prospective customers. This conversation is worth considering for legal, medical, electrical, HVAC, and other work where permission matters more than a headline price. Start with one question your team can answer clearly, not a page that tries to cover every rule.
9. Put your numbers beside the “is it worth it?” answer
“Is spray foam insulation worth it for a 1,800 sq ft house?” or “does Invisalign actually work at my age?” is a request for help deciding whether to spend. A general answer can frame the costs and benefits. Your ad should offer the numbers that matter to the particular service you sell.
Offer a savings or payback estimate, not a free consult. The landing page needs one simple input, your installed price range, and two before-and-after numbers you can stand behind. Give the buyer a way to judge the purchase, then a way to book. This is not the first conversation I would buy on a small budget: these leads can take weeks rather than hours, and judging the page on day three will tell you very little.
That timing is why I would test only two or three of the nine with a $1,500 to $3,000 budget. Start with No. 1, the provider shortlist; add No. 2, emergency, if you can answer the phone; add No. 3, availability, if your calendar and follow-up can support it. Give each conversation its own offer and page. Set a pause rule before launch: after 20 to 30 clicks with no qualified calls or booked estimates, stop that test and put the money toward the stronger one. Early ChatGPT ad results vary, so treat this as a controlled test rather than a settled playbook.
Each Monday, ask for spend, qualified calls, jobs booked, and cost per booked job by conversation type. Not a screenshot of impressions. If one test spends $800 for 11 qualified calls and four booked jobs, while another spends $800 for 22 calls and one booked job, the second loses despite its cheaper cost per lead. That is the report your budget needs.
If you hire help, ask who will build the distinct pages, rewrite the context hints, and stop a weak test. A generic landing page and monthly PDF do not answer those questions. For this work, I would put groas’s flat-fee Search Ads management across Google, Bing, and ChatGPT, with logged actions and a human owner, ahead of a freelancer juggling accounts or a traditional percentage-of-spend retainer. The work is in the daily decisions, not the dashboard tour.
I would still start with No. 1: the provider shortlist. The buyer has decided to call someone; your page has to give them a reason to call you. Build that page first, then add emergency and availability only if your operations can deliver what those ads promise. One conversation, one page, one booked-job number to judge it by.

