Case study · The Comfort Authority · HVAC · Tampa, FL
What happens if you stop advertising in two thirds of your service area? Spoiler: fewer calls, and more of them turn into paid work
- Cost per sold job
- −30 to −35%
- Lower is better
- 90 days against the 90 before · Call tracking
- Jobs sold from paid search
- 1.6×
- Higher is better
- 90 days against the 90 before · Client CRM
- Calls from ads
- −22%
- Lower is better
- 90 days against the 90 before · Call tracking
On this page
At a glance
Twenty-two per cent fewer calls, and the cost of a sold job came down by about a third.
Cost per sold job: 90 days against the 90 before · Call tracking · All three results
- Company
- The Comfort Authority
- Trades
- Air conditioning, Heating, Plumbing
- Market
- Tampa, FL
- Trading since
- 2014
- Licences
- CAC1818057, CFC1433906
- What we run
- Google Ads
- Local Services Ads
- Call tracking
- Landing pages
- Window
- 90 days against the 90 before them
Who this is
The Comfort Authority has been fixing air conditioning and plumbing in Tampa since 2014, out of one shop on North Nebraska Avenue. Two licences in the footer, CAC1818057 for the mechanical side and CFC1433906 for the plumbing side. Family owned, Steve and Gerri Cox.
Their whole pitch is that most contractors sell cooling and call it air conditioning. Real air conditioning, they argue, is humidity, ventilation and air movement together. In Florida that is not a slogan, it is the actual job.
The number nobody had looked at
Their service area page lists forty-seven cities. Apollo Beach, Clearwater Beach, New Port Richey, Plant City, St Pete Beach, Wesley Chapel. From North Nebraska Avenue, the far end of that list is over an hour each way in season, across the Howard Frankland.
The ad account did not know that. It bought a click in Clearwater Beach at the same price as a click in Temple Terrace, ten minutes up the road. Both looked identical in the reporting, because the reporting stopped at the call.
| Before | After | |
|---|---|---|
| Under 15 minutes | 41 | 100 |
| 15 to 35 minutes | 74 | 62 |
| Over 35 minutes | 100 | 11 |
What we were afraid of
Cutting two thirds of a map in May, in Tampa, is not a comfortable thing to propose. If it went wrong it would go wrong in the twelve weeks that carry the year.
So we did not cut anything until the measurement could survive the argument. First a call could be followed all the way to a sold job. Only then did anybody touch a radius.
What the account was actually buying
With calls tied to outcomes, the search terms sorted themselves. A meaningful share of the spend was going to people who were never going to hire a contractor: homeowners fixing it themselves, technicians looking for work, shoppers pricing window units.
| Search term | Decision | Why |
|---|---|---|
| ac repair near me | Kept | The term the business runs on. Kept it, capped the radius so it stopped buying Clearwater. |
| how to reset ac breaker | Cut | Somebody fixing it themselves. Costs a click, never becomes a call. |
| hvac jobs tampa | Cut | A technician looking for work, landing in a repair campaign. |
| portable air conditioner | Cut | A product search. Nobody calls a contractor for this. |
| emergency plumber tampa | Moved | Real intent in the wrong campaign, competing with HVAC for the same budget. |
The order things happened in
None of this is clever on its own. What matters is the sequence: nothing was scaled before the measurement was trusted, and nothing was cut before there was somewhere better to put the money.
- Weeks 1–2 Calls wired through to the CRM, so a call could be followed to a sold job rather than counted at pickup.
- Week 3 Radius rebuilt on drive time instead of city names. Forty-seven cities became a shape a van can finish.
- Weeks 4–6 Search terms cleared out. Three thin campaigns collapsed into one with enough budget to learn.
- Week 7 Plumbing given its own budget instead of losing every auction to air conditioning.
- Weeks 8–12 Spend moved onto the hours a person actually answers the phone.
What came back
Fewer calls. That was the point, and it is still the sentence that makes owners nervous.
The calls that remained were closer, and closer calls book better and close better. The whole gain is in the last two rows: the same funnel, narrower at the top and fatter at the bottom.
| Before | After | Change | |
|---|---|---|---|
| Calls from ads | 100 | 78 | −22% |
| Booked visits | 44 | 51 | |
| Jobs sold | 19 | 31 | +63% |
Where the season put a ceiling on it
Tampa does not have a shoulder season the way Dallas does. Cooling runs most of the year and the crew is at capacity from May to September, which means the ceiling on this account is not the budget, it is the calendar.
Past a point, more calls in July are not worth buying. They are worth moving to October.
| Month | Index (busiest month = 100) |
|---|---|
| January | 38 |
| February | 41 |
| March | 52 |
| April | 68 |
| May | 88 |
| June | 100 |
| July | 97 |
| August | 94 |
| September | 86 |
| October | 63 |
| November | 44 |
| December | 40 |
Results
- Cost per sold job
- −30 to −35%
- Lower is better
- 90 days against the 90 before · Call tracking
- Jobs sold from paid search
- 1.6×
- Higher is better
- 90 days against the 90 before · Client CRM
- Calls from ads
- −22%
- Lower is better
- 90 days against the 90 before · Call tracking
What this does not prove
- The two windows are not weather-matched. Two Tampa quarters are never identical, and some of this movement belongs to the season rather than to us.
- Jobs sold come from the client's CRM, which means they depend on technicians closing tickets correctly. We did not audit that, we read it.
- The plumbing side is too small a sample over this period to say anything honest about on its own, so it is not broken out.
- Nothing here separates the effect of the radius change from the effect of the search term cleanup. They shipped three weeks apart in the same account and we did not hold one back to prove the other.
- We cut the far cities. We did not prove they are unprofitable at every budget — only that they were unprofitable at this one, against the near ones.
What to take from this
- Measure the drive, not the click Two clicks at the same price can differ by two hours of a technician's day, and nothing in an ads platform will tell you which is which.
- Count the sold job or do not count A booked call is a promise. Optimising toward promises is how an account ends up buying the cheapest possible way to be disappointed.
- Give plumbing its own budget In a shared budget the higher-ticket trade wins every auction, and the other one quietly stops existing without anybody deciding that.
- Buy the hours somebody answers Spend that lands when the phone rolls to voicemail is not cheaper traffic, it is a full-price lead handed to whoever the caller rings second.
- In a capped season, move demand instead of adding it When the crew is full in July, another July call is worth less than the same call in October, and the account should be priced that way.
Questions people ask about this one
- Does cutting the service area mean losing customers?
- It means declining a share of them on purpose. A job ninety minutes away occupies a technician for most of a day, closes at a lower rate and pushes back the jobs nearby. Cutting that work costs real calls, and the trade is that the remaining hours go to work that pays better per hour. We would not recommend it to a company with idle crews.
- Why did calls go down but sold jobs go up?
- Because calls and sold jobs are different things and only one of them pays a technician. Removing search terms that never become jobs, and areas a van cannot reach profitably, removes calls. What is left books at a higher rate and closes at a higher rate, so fewer calls carry more sold work.
- How do you know a call turned into a paid job?
- The call is tracked from the ad through to the CRM record, so the sold ticket carries the campaign that produced it. Without that link an ads account can only optimise toward calls, and calls are not the thing a contractor sells.
- Would this work for a company with more than one location?
- The method is the same but the map is not. With two shops the radius is drawn around each of them and the overlap has to be resolved deliberately, or both locations bid against each other for the same household. A single-location company is the simpler version of this problem.
