A twelve mile radius buys a different job at ten in the morning than at five
Distance based targeting quietly sells arrival times your dispatcher cannot keep.
Twelve miles inside the Perimeter at ten in the morning is routine. The same twelve miles at five, crossing I-285 or dropping down a radial interstate, is a different afternoon entirely.
Radius targeting has no idea. It draws a circle and buys clicks from every part of it at every hour of the day.
Build geography from ZIP codes or municipalities grouped by how long the drive genuinely takes at the hours you run. Then use ad schedule adjustments on the edges: the core runs all day, the far edges bid up mid morning and pull back before the afternoon crush.
The test is unglamorous. If jobs booked from the outer edge keep turning into late arrivals or reschedules, you paid for that click twice.
Split the account at the Perimeter before you split it by service
Inside and outside the Perimeter behave like separate markets, and one campaign averages them into something you cannot act on.
Intown work leans older housing, tighter access and street parking. A newer subdivision in Johns Creek or Alpharetta produces different jobs, different tickets and different close rates.
Separate campaigns give you separate budgets, separate bids and separate reporting. You can see that Decatur is cheap and closes well, or that a north side campaign eats budget on calls that never book.
Do not shatter it further than the data supports. Two or three geographic groupings is usually the ceiling for one metro, and a fourth added for tidiness just slows every decision down.
Name the campaigns the way your team talks: inside the Perimeter, north OTP, east side. Reporting people can read is reporting people act on.
Burst pipe money and water heater money should not share a daily budget
Emergency demand and planned demand want different bids, different pages and different pacing, so they need different campaigns.
An emergency campaign should be allowed to spend. Someone with a live leak converts fast, tolerates a premium and does not shop three quotes.
Planned work behaves the opposite way. A water heater replacement or a repipe gets compared, delayed and quoted twice before anyone signs.
Put them in one campaign and the emergency terms will eat the budget on a bad weather day while the planned terms never get a fair run. Split them, then match the landing page: one built around a phone number and an arrival window, one built around explanation and a quote form.
If you run Local Services Ads, remember they compete for the same emergency intent. Budget and report them next to the emergency campaign, not next to planned work, or the numbers will flatter the wrong thing.
Cobb clicks and DeKalb clicks close differently, so stop averaging them
Cost per lead is a county level number pretending to be a metro level one.
Two clicks at the same price can be worth wildly different amounts. One becomes a booked job at a healthy ticket. The other becomes a quote for a landlord collecting three of them.
Track cost per booked job and break it out by county. Fulton, DeKalb, Cobb and Gwinnett will not perform alike, and the differences are usually stable enough to reallocate against.
Feed booked jobs back into the platform as your primary conversion where the setup allows, rather than form fills and raw call counts. Bidding improves when the signal sits closer to money.
No agency can promise a cost per booked job in advance. What you can insist on is that the number gets calculated at all, and that it is split by geography before anyone calls a campaign a success.
One hard freeze can spend a January budget in two days
Two demand peaks a year make flat monthly pacing the wrong default.
A freeze in the forecast turns an ordinary week into the busiest days of the quarter. Searches for burst pipes, frozen lines and no water arrive all at once.
A rigid daily budget caps you on exactly the day the market is least price sensitive. Room in the budget, plus someone actually watching, captures work that carries a slow February.
The reverse holds too. Do not run a peak season pace through a quiet shoulder stretch just because the calendar says a month is a month.
Decide the pacing rule before the season starts: which campaigns get released, who has authority to lift a cap, and what the ceiling is. Making that call at seven in the morning with three phones ringing rarely goes well.
What to refuse: renters, price shoppers and people hiring a plumber
Most of the money wasted in a plumbing account goes to searchers who were never going to become customers.
Start with intent mismatch. DIY, how-to, diagrams, parts and fittings, code lookups. Add cheap and free phrasing if you are not competing on price, because those clicks convert into arguments rather than jobs.
Add the employment layer. Plumber salary, apprenticeship, hiring, license requirements. Those searches are common, they look commercial to the algorithm, and they cost exactly what a real one costs.
Add the wrong buyer. Renters looking for a landlord's obligation, home warranty administrators, competitor brand names you cannot service under, and commercial or municipal work if you do not do it.
Handle geography with exclusions rather than bid adjustments alone. An outer county you cannot reasonably reach should be excluded outright, because a downward adjustment still lets some through, and a call you have to refuse still costs the click and the minute.
Questions we actually get
- Should we run Local Services Ads instead of search ads?
- Most plumbing accounts end up running both, because they serve different moments. Local Services Ads compete for immediate, high urgency intent. Search campaigns carry the considered work like repipes and sewer replacement, where a landing page has to do real explaining. Budget and measure them separately or the emergency side will make the planned side look worse than it is.
- What is a reasonable budget to start with?
- It depends on your ticket, your close rate and how much of the metro you actually serve, so any figure quoted before those are known is invented. The more useful starting question is how many trucks you can fill next week. Buy demand for that capacity first, then expand geography once the cost per booked job is known and stable.
- Why measure cost per booked job instead of cost per lead?
- Because leads are not evenly valuable in this trade. A form fill from a landlord shopping three quotes and a call from a homeowner with water on the floor cost the same and are worth nothing alike. Cost per booked job survives a slow week and a bad lead mix; cost per lead does not.
- Can we just target a radius around our shop?
- You can, and it will overspend on the parts of the circle you cannot serve at the hours that matter. Grouping ZIP codes by realistic drive time takes an afternoon and stops the account from selling arrival windows the schedule cannot honor.
- How quickly should we expect the account to settle?
- Nobody can promise a timeline, and account performance moves with weather and season here more than most trades. What is reasonable to expect is a defined structure, a negative keyword list that grows every week from the search terms report, and reporting split by county from the first month rather than the sixth.
