TL;DR
- AI search leads read differently at intake. The customer has decided they need the service and is asking who to hire, not whether.
- A marketplace sends one form to four or five companies. An AI answer names one to three, so the lead often calls only you.
- Ticket size follows the trade, not the channel. The channel changes how many leads at that ticket actually book.
- The proof sits in your own intake log. Add one free-text question, how did you find us, and read the words.
- Across 35 sales calls I ran between March and August 2026, price came up once. Lead quality came up on nearly all of them.
Why the Log, Not the Adjective
Adjectives settle nothing. The log does.
An intake log needs three columns to answer this: what the customer said when you asked how they found you, what they booked, and what the invoice came to. Sort by source. Read. Ten minutes.
I ran 35 growth analysis calls with local business owners between March and August 2026. Price came up once. The question on nearly every call was a version of this: will these be real jobs, or the same people who message six companies on an app and take the cheapest quote? The log answers it, trade by trade.
What the Two Leads Look Like at Intake
Put the two intake notes side by side and the difference shows before anyone mentions a ranking. A marketplace lead tapped a category and filled a form that went to four or five companies at once. An AI search lead typed a full sentence into ChatGPT, Gemini or a Google AI Overview, got an answer naming one to three businesses with a reason each, and rang one of them.
| Signal | Marketplace lead | AI search lead |
|---|---|---|
| How they found you | Tapped a category, got matched by the platform | Asked a question in a full sentence, got named in the answer |
| Companies contacted | Four or five at once, by design | One or two, because the answer listed one or two |
| First question on the call | How much? | Can you do it this week? Do you handle my situation? |
| What they already decided | That they want quotes | That they need the job done |
| The intake note reads like | A price check | A brief |
| What you compete on | Price, against four others | Fit, against nobody or one other |
The ticket doesn't change with the door. An implant costs what an implant costs. What changes is how many of those leads become a booked chair, because the customer arrived with a shortlist of one.
What the Log Shows by Trade
These are the shapes of intake notes across the trades I work in, paraphrased so nobody is identifiable. The ticket column is the industry range, so you can check it against your own price list. One accounting client of mine averages 10 to 20 extra calls a month across Google and ChatGPT combined. Same pattern every time: the caller has decided and is asking who.
| Trade | What they typed | What they booked | Typical ticket |
|---|---|---|---|
| Dental | Dentist near me who does implants and takes my insurance | Consult, then the implant | $3,000 to $5,000 per implant |
| HVAC | AC stopped cooling, who can come out today | Same-day repair, often a replacement conversation | $300 to $800 repair, $6,000 to $12,000 replacement |
| Plumbing | Licensed plumber for a leaking water heater | Water heater replacement | $1,500 to $3,500 |
| Restoration | Water damage company that works with insurance | Mitigation job, often billed to the carrier | $2,500 to $7,500, higher with insurance |
| Cleaning | Deep clean before move-out, available this weekend | One-off deep clean, sometimes recurring | $250 to $600 |
| Bookkeeping | Bookkeeper for a small business who does catch-up work | Monthly engagement plus catch-up | $300 to $800 per month |
Read the second column again. Not one of those is a price question. Each names a situation, a constraint, or both. That's the tell. A price shopper types cheap plumber. A customer with a leaking water heater types the leak.
Why AI Search Leads Read Differently
I say this on nearly every sales call. By the time someone asks ChatGPT who they should hire, they've already decided they need a dentist. They're asking who, not if.
A marketplace puts price in front of the customer before any company has spoken, and hands the same form to five of them. The customer becomes a price arbiter because the platform made them one. An AI answer names a few businesses and gives a reason for each: takes this insurance, handles emergency calls. The customer arrives holding that reason and often repeats it to your front desk word for word.
That makes attribution easier here than on any other channel. My own first booked call sourced from ChatGPT landed at the end of August 2026. I didn't find it in a dashboard. The prospect typed where they found me into the booking form. Your customers will do the same if you give them a box.
How to Read Your Own Log
You don't need us for this part. Four changes to intake, then one read a month.
- Add a free-text question to every form: how did you find us? No dropdown. A dropdown files ChatGPT under Google or Other, and the sentence is the whole point.
- Ask the same question on the phone and have the front desk write the sentence, not a tag. Asked ChatGPT for a plumber who works with insurance is data. Online is not.
- Tag the source afterwards, from the words. Three tags cover it: marketplace, Google search, AI search.
- Once a month, three numbers per tag: leads in, jobs booked, dollars invoiced. That's the report. Anything longer is hiding something.
If you've hired an agency for this, those three numbers should exist by day 30. I wrote up what should be visible by day 30 so you can hold them to it.
Which Bucket Are You In?
Three buckets. Pick the one that matches your log, not your hopes.
Build for AI search now if
- Your average ticket is over $1,000. Dental, HVAC, plumbing, restoration, accounting.
- Your customers ask who questions: who takes my insurance, who comes out today, who does catch-up.
- Your intake already captures the sentence, so you'll see the first AI lead the week it lands.
Read which local businesses should invest in AI search first.
Keep the marketplace and add AI on the side if
- You need volume this month and can live with being one of five quotes.
- Your ticket is under $300 and the margin survives a shared lead.
- You can run both for a quarter and compare the three numbers by tag.
See AI search vs Google Ads for restoration companies for two channels side by side.
Choose neither yet if
- Your intake log doesn't exist. Fix that first. A channel you can't read is a channel you can't judge, and you'll end up where you were with the last agency.
Start with how to set up the next agency so you can fire it with evidence.
What Each Lead Costs You
A shared marketplace lead runs $15 to $150 depending on the trade, charged whether or not you win. At a one-in-five close, that's $75 to $750 per booked job before any labor.
An AI search retainer for one location runs $800 to $3,000 a month. At $2,000, one HVAC replacement or two restoration jobs covers it. What sits inside that range is in how much AI search optimization costs. Before money moves, get the deliverables onto a one-page scope, so the log has something to be measured against.
Frequently Asked Questions
Do leads from ChatGPT and AI search turn into real jobs, or are they price shoppers?
In the intake logs I read, they book at a higher rate than marketplace leads from the same trade, and the reason sits in the customer's own words. A marketplace lead asked for quotes and got five companies. An AI search lead asked a question, got an answer naming one to three businesses, and called one. They've decided they need the service. They're asking who. The ticket is set by the trade, so the difference shows up in close rate and in how many quotes you're up against, not in the invoice. Read your own log by source and you'll see the same split.
How do I tell an AI search lead from a marketplace lead in my intake log?
By the words, not by a tag. Add a free-text question to every form and phone script: how did you find us? Have the front desk write the sentence the customer says. An AI search lead tends to repeat the reason the answer gave them, such as takes my insurance or handles emergency calls. A marketplace lead often can't tell you which company they're calling, because one form went out and four companies rang back. Tag the source afterwards from the sentence, using three tags: marketplace, Google search, AI search. Count leads, bookings and invoiced dollars per tag once a month.
Why do AI search leads contact fewer companies than marketplace leads?
Because the answer they read was short. A marketplace is built to distribute one request to several providers, so the customer expects to compare and the platform expects them to. ChatGPT, Gemini and Google AI Overviews return a handful of names with a sentence of reasoning each. The customer picks the one whose reason matches their situation and calls it. In practice that means you're often the only call, or one of two. The competition happened inside the answer before the phone rang, which is why the intake note reads like a brief rather than a price check.
Which trades see the largest tickets from AI search leads?
The ticket follows the trade. Dental implants run $3,000 to $5,000 each. An HVAC replacement runs $6,000 to $12,000, with repairs at $300 to $800. A water heater replacement sits around $1,500 to $3,500. Water damage mitigation runs $2,500 to $7,500 and climbs when insurance is involved. Monthly bookkeeping lands at $300 to $800 per month, which compounds over years. Cleaning and chiropractic carry smaller tickets, $250 to $600 per deep clean and $60 to $150 per visit, and lean on repeat business. AI search doesn't raise these numbers. It changes how many of the leads at each ticket actually book.
How much does AI search optimization cost compared with buying marketplace leads?
A shared marketplace lead costs $15 to $150 depending on the trade, and you pay whether or not you win the job. At a one-in-five close rate that's $75 to $750 per booked job before any labor. An AI search retainer for a single location runs $800 to $3,000 per month, covering profile work, content, and intake tracking. At $2,000 a month, one HVAC replacement or two restoration jobs covers the fee. For a cleaner at a $300 ticket the math needs volume, so run both channels for a quarter and compare the three numbers per source before committing.
How long before AI search leads show up in the intake log?
Expect the first ones inside 30 to 60 days if the profile and content work is done in the first fortnight and your intake captures the sentence. If your form has no free-text field, they may already be arriving and getting filed under Google or Other. Fix that first. By day 30 you should be able to point to at least one intake note that names an AI tool in the customer's words. By day 90 you should have enough per source to compare close rates. If an agency can't show you either of those, that tells you about the agency, not about the channel.
Should I stop buying marketplace leads once AI search leads arrive?
Not on day one. Run both for a quarter with the same three columns per source: leads in, jobs booked, dollars invoiced. Then look at cost per booked job, not cost per lead. Marketplace leads look cheap per lead and expensive per job, because you win one in five and pay for all five. AI search looks expensive per month and cheap per job once the log fills up. Most owners I work with cut marketplace spend gradually as the AI column grows, rather than switching it off. The log tells you when. Stopping before you can read the log is guessing.
What should I ask an agency about lead quality before I pay?
Four questions. First, how will I see the customer's own words at intake, and who sets that up? Second, what three numbers will you report each week, and will one of them be dollars invoiced? Third, what should be visible in my log by day 30, in writing? Fourth, can you show me an intake note from a current client that names an AI tool, with the name removed? An agency selling rankings will stumble on all four. An agency selling jobs will have the answers ready, because the log is what it gets judged on. Get those answers onto a one-page scope before any money moves.
Want your log read?
Book a growth analysis call. I'll pull your intake by source, show you what booked and what invoiced, and tell you plainly whether AI search is worth your money this quarter.
Book a growth analysis callNo pitch deck. If the log says the marketplace is winning for you, I'll say so. Pick a time here.
About the author
Matthew Johnson is the founder of Pleiades Consultancy. He previously scaled his own marketing agency to multiple six figures before serving as CMO of an Amazon agency, where the client base tripled from 15 to 45 active clients during his tenure. He worked with some of the largest names in e-commerce, including Ridge Wallet, HexClad, BK Beauty, The Woobles, Walkize, Lonely Planet, and Obvi. He now works with local businesses to maximize their client acquisition and visibility through AI search with ChatGPT, Claude, Gemini, Perplexity, and Bing Copilot.