Marcus is made up, but you could find his business in most Florida suburbs. He runs three trucks out of a unit in Brandon, just east of Tampa, doing mostly termites, rodents, and the endless ant calls that come every rainy season. Last spring he was spending more on Google Ads than he ever had, and his phone wasn’t ringing any more often.
His nephew, who “does computers,” suggested he sit down one Sunday and listen to a month of recorded calls, then look at what people had actually searched for before landing on his site. Marcus did it with a notepad and a large iced tea. What he found surprised him.
Most of it came down to the gap between a search and a booked visit. A good pestcontrol call engine closes that gap. A relevant search lands on a page about that exact pest, the phone number is easy to tap, and someone answers. Marcus had pieces of that. He was missing some of the joins.
The termite people were landing on the ant page
Half his ad clicks went to the homepage, a long scroll covering every bug he’d ever treated. A homeowner who’d just found tiny wings on a windowsill had to wade through cockroach and mosquito paragraphs before finding anything about termites. A fair number gave up before they got there.
He made a separate page for each main service. The termite page explains what an inspection involves, how long it takes, and what happens if he finds activity. The rodent page talks about sealing entry points, not just setting traps, because that’s the question callers ask most.
This now matters for AI search too. When a homeowner asks ChatGPT or Gemini who does termite inspections near Brandon, those tools look for pages that clearly say what a business does and where it works. A homepage that says “we handle all your pest needs” gives them very little to quote.
Ads he was paying for that were never going to work
The search terms report was the part that made Marcus wince. He’d paid for clicks from “pest control jobs Tampa,” “how to become an exterminator,” and someone asking how to kill fire ants with vinegar.
None of those people were ever going to book a $150 inspection. Adding them as negative keywords freed up a chunk of his budget almost overnight. He also linked each ad to its matching page. The rodent ad now opens the rodent page, not a contact form that’s lost all the context from the search.
He also changed his ad schedule. He’d been running “Call now!” ads at 9 p.m., when nobody was there to pick up.
The call that rang eleven times
This was the hard part to listen to. His call-tracking software, the kind that records calls and uses AI to transcribe and tag them, showed that a lot of “calls” never reached a person.
Some were just taps on the phone button that got abandoned before connecting. Others rang out while his office manager was on another line. One woman with a rat in her kitchen let it ring eleven times on a Saturday morning and then called someone else. The transcript tags made it easy to search for every call that said “rat” or “termite” and check what happened next.
His old marketing report had counted all of those as leads. In reality, a tap on a phone number and a conversation with a customer are two different things.
What happens after hours now
Marcus didn’t want to hire someone for nights and weekends, so he tried an AI phone assistant and a chat widget on the site. They’re set up to do a small number of things well. They confirm which pests he treats and which towns he covers, take the caller’s details and the problem, and promise a callback by 9 a.m.
He was careful about the limits. The assistant doesn’t quote prices, diagnose anything, or book specific time slots, because the one time it tried, it promised a Sunday visit when nobody was working. Businesses are going much further with AI marketing agents these days, but for a three-truck operation, narrow and reliable beat clever.
The website copy he almost let a chatbot write
When he redid the service pages, Marcus’s first instinct was to have an AI tool write them. The drafts were tidy and completely generic. Every one promised “comprehensive solutions” and “peace of mind.”
In the end, he used the AI to tidy up his own notes and spent most of the time on the details. He wrote about how long a termite inspection really takes, what customers need to move out of the garage beforehand, and why he’d sometimes recommend sealing gaps instead of a second spray. A few habits for checking AI-generated content helped him catch two claims the tool had invented about treatments he doesn’t offer.
He also stopped polishing reviews. Real customer comments go up as written, typos included. Fake or AI-generated reviews have been squarely in the FTC’s sights since its 2024 rule, and real ones mention the details people care about anyway. Customers kept praising his technician Luis for explaining things clearly, so “clear explanations” went into the service pages too.
The form had seven fields
On his phone, the inquiry form asked for name, email, phone, address, pest type from a dropdown, “severity,” and a preferred date. Nobody knows how severe their termite problem is. That’s what they’re paying him to find out.
He cut it down to name, phone, the pest they think they have and a short box to describe it. Underneath, he added a line saying what happens next: “We’ll call you back within two business hours.” His office really can manage that.
The monthly ten
Now, at the start of each month, Marcus pulls ten random inquiries and follows each one to the end. Did it become an inspection, a quote that was turned down, a job, or a call about bed bugs, which he doesn’t treat?
It changed how he spends money. One campaign brought in lots of cheap inquiries, but most were for services he doesn’t offer, so he turned it down. Another brought fewer calls, but nearly all of them booked, so he moved budget there.
Last month, two of the ten were missed calls nobody returned. Before buying more clicks, he fixed that by giving his office manager a callback list that has to be cleared before lunch.
