Draw the geography on tollway drive times, not on the Metroplex
A radius that looks sensible on a map buys clicks from places your trucks will not reach at a profit.
Twenty five miles from a shop in Addison touches a lot of ground on paper and very little of it at five in the afternoon. Traffic, not distance, decides what a crew can serve, and this metro is enormous and freeway dependent.
Start from evidence. Pull the addresses of every job you ran last year, plot them, and look at where the profitable ones cluster. Most shops find a tight core, a productive ring, and a scatter of one-off jobs that cost a half day of drive time each.
Target by ZIP or by city rather than by radius, so the edge of your map is a decision rather than an accident. Exclude the places you turn down, including the ones that keep calling.
Set location targeting to people in or regularly in your area rather than people showing interest. Otherwise you pay for a landlord in another state researching a rental, and for anyone reading about Dallas from anywhere else.
A no-cool call and a five ton changeout are two different auctions
Repair intent and replacement intent behave differently enough that they should never share a budget, a bid or an ad.
Repair searches are urgent and shallow. The person converts on a phone call within minutes, compares almost nothing, and rewards whoever looks reachable right now. Ad copy should promise an answer, not a discount, and call assets matter more than landing page polish.
Replacement is slower and more expensive per click. The buyer is comparing bids, thinking about financing, and often bringing a spouse into it. Those clicks deserve a landing page about what a quote includes and what changes the price.
Keep them in separate campaigns so you can pace them separately and read them separately. Blend them and your cost per lead looks fine while your cost per sold changeout quietly rises.
Two more belong in their own structures. Maintenance plan campaigns run in spring and fall when nothing is broken. Light commercial should never sit inside the residential account, because the buyer, the budget and the sales cycle are all different.
Local Services Ads sit alongside all of this. They charge per lead, they lean on license and insurance verification you already keep current, and they are worth running as their own line item with their own cost per booked job.
People checking an Oncor outage map are not customers
Most of the waste in an HVAC account is search terms that look relevant and are not.
After a storm, searches about power, outages and restoration surge across North Texas. Broad match will happily put you in front of every one of them. Add those terms as negatives before the next storm season rather than after.
The rest of the usual list: model and part numbers, DIY repair questions, capacitor and refrigerant pricing, window units, portable units, mini split kits sold online, technician jobs, HVAC schools and apprenticeships, and rental or tenant searches if you do not serve landlords.
Read the search terms report weekly for the first few months of any account, then monthly. Build a shared negative list applied at the account level, with campaign level lists for the distinctions that matter, like keeping commercial terms out of the residential campaign.
Competitor names deserve a judgment call rather than a rule. They can be cheap and effective in a market with steady relocation, since newcomers hear a name at a barbecue and search it. They can also be a way to donate money to a franchise with a bigger budget. Test it with a small cap and read the booked jobs, not the clicks.
The first hundred degree stretch sets your budget, not the calendar month
Demand here arrives in bursts, so the account should follow the weather rather than an even monthly plan.
Extreme summers and the occasional hard freeze both produce days where every phone in the metro rings at once. Costs rise on those days because every competitor bids harder. The temptation is to pull back. The better question is whether you can still dispatch.
If the board is full and the phone is going to voicemail, extra spend buys you bad reviews. Pause or cap. If you have capacity, those are the days worth spending into, because urgency shortens the sale to a single call.
Spring and fall are cheaper and slower, which makes them the right window for replacement research and maintenance plans. A shop that only spends in July trains itself to buy the most expensive clicks of the year.
Set budgets against dispatch capacity for the coming week rather than against a monthly figure. Give someone the job of adjusting them, and give them the schedule board to look at while they do it.
Bidding Fort Worth from a Dallas shop is a second business, not a wider radius
Dallas and Fort Worth are separate markets with separate downtowns and an hour of freeway between them.
Widening a Dallas campaign to include Tarrant County usually produces the same result: the algorithm finds the cheapest clicks, which come from wherever the competition is thinnest, and the jobs land at the end of a long drive.
If Fort Worth is a real target, run it as its own campaign with its own budget, its own ad copy naming Tarrant County towns, and its own landing page. A local phone number helps. So does a genuine presence.
Expect different numbers. Cost per click, close rate and average ticket all tend to differ across the two anchor cities, and you will only see it if the reporting keeps them apart.
If no truck lives west of Arlington, do not bid it at all. Buying inquiries you cannot serve profitably is more expensive than not appearing.
Google optimizes for whatever you feed it, so feed it booked jobs
Smart bidding against raw form fills will reliably find you the cheapest form fills, which are not what you sell.
Send conversions back from the field software. At minimum, mark which inquiries became booked jobs. Better, import the sold value of the changeout so bidding learns which searches produce revenue rather than which produce contact.
Pick a stage the office actually updates. An offline conversion import that depends on a field nobody fills in will make the account worse, not better, because it will teach the system that most of your good leads were failures.
Report two numbers to the owner every month: cost per booked job and cost per sold replacement, split by county. Watch them diverge. A campaign can look efficient on booked repair calls while producing almost no replacement work.
Hold the account to a decision rule set in advance. Which campaigns get more money, which get paused, and at what point a market like Denton County is judged. Deciding that in advance stops the monthly conversation from turning into a debate about whether the season was unusual.
Questions we actually get
- How much should a Dallas HVAC company spend on Google Ads?
- Work backward from capacity rather than from a benchmark. Decide how many additional jobs per week you can dispatch, estimate what a booked job is worth to you, and cap spend where the math stops working. Any agency quoting a budget before seeing your close rate and your average ticket is guessing.
- Are Local Services Ads worth running alongside search?
- For most residential HVAC shops they are worth testing, since they charge per lead and lean on license and insurance verification you already maintain. Run them as their own line item with their own cost per booked job, and dispute bad leads promptly. Treat them as an addition to search, not a replacement.
- Why does our cost per lead jump every July?
- Because every competitor in the Metroplex is bidding into the same heat wave. Auction prices rise with demand. The useful response is to check whether you can still dispatch what you buy, spend into the days you have capacity, and stop judging the account on a month when the weather is doing most of the work.
- Should we bid on competitor brand names?
- Sometimes. In a market with sustained relocation into Plano, Frisco and McKinney, people search a name they overheard without loyalty attached, and those clicks can be inexpensive. They can also be a slow donation to a larger advertiser. Cap it small, run it separately, and judge it on booked jobs alone.
- Do we need separate campaigns for Fort Worth?
- If you genuinely serve Tarrant County, yes. Separate budget, separate copy, separate landing page, separate reporting. Merged into a Dallas campaign, the spend drifts toward whichever side is cheaper that week and the results become impossible to read.
