Radius rings buy clicks from the far side of I-285 at four in the afternoon
A circle drawn around your yard is a promise your dispatcher cannot keep during the hours homeowners want appointments.
Distance is the wrong unit here. Twenty miles north at ten in the morning is a comfortable trip. The same twenty miles at four means an estimator loses the rest of the day, and the homeowner is annoyed before the tape measure comes out.
Build the targeting from named cities and ZIP codes grouped into drive-time bands from the yard, not from a mileage ring. Bid the close band hardest, the middle band at a discount, and leave the far band off until a route already puts a crew nearby.
Set location targeting to people in your area rather than people merely interested in it. The looser setting quietly buys clicks from out-of-state owners and researchers.
Dayparting is worth testing against the same logic. If nobody can service an appointment request that lands at five, ask whether the far band should be bidding at all in the late afternoon.
Patio and entry doors run on a different clock than a whole-house window quote
Door buyers usually have a date in mind and window buyers usually have a budget conversation ahead of them, so the two should never share a budget line.
Something forced the door search. Rot at the threshold, a slider that will not lock, a dog that destroyed the screen, a holiday coming. The homeowner wants it handled and decides quickly.
Whole-house window work is a considered purchase. Multiple quotes, a spouse, sometimes financing, and weeks between the measure and the signature.
Blend them and the door campaign's fast conversions make the account look healthy while the window campaign starves. Separate campaigns, separate budgets, separate landing pages, separate targets.
Keyword themes should follow the same line. Door searches name the door. Window searches name a symptom, a count or a room. Writing one ad for both produces copy that speaks to neither.
Rental turnovers, builder bids and DIY sizing should be spending somebody else's budget
The search terms report, not the keyword list, is where the money actually goes in this trade.
The predictable drains are glass-only repair, screen work, window cleaning, tinting and anything with how to measure or how to install in it. None of those people are buying an installed unit from you.
Two more are specific to a metro this size. Apartment and rental turnover shopping, where the buyer wants the cheapest possible unit and a volume price, and new-construction or builder bid requests, which run on a different sales process entirely. Decide deliberately whether you want either, then negatives accordingly.
Review the terms weekly for the first two months, then monthly. Broad match with automated bidding drifts, and it drifts fastest in metros with lots of adjacent commercial language.
On Performance Max and Display, exclude placements as aggressively as you exclude terms, and keep brand terms in their own campaign so the automation cannot claim credit for people who already knew your name.
The budget should climb with the first cold snap and again with the spring remodel
Two genuine seasons mean two ramps a year, and a flat monthly cap wastes money in the trough and runs dry in the peak.
Cold weather produces urgency that no ad copy can manufacture. Drafts, condensation on the inside of the glass, a room nobody sits in. Demand rises quickly and falls back.
The spring wave is calmer and larger. Appearance, noise, doors, remodels that were priced over the winter. It rewards the accounts that were already running rather than the ones switched on in April.
Pace against installed capacity, not the calendar. When the backlog runs past what you are comfortable telling a caller on the phone, pull spend back rather than buying inquiries that will wait and then cancel.
Keep a floor in the slow weeks. Accounts that go dark lose their footing and cost more to restart than the savings were worth.
Local Services Ads and search ads compete for the same phone, and only one of them shows a star rating
Two Google products can bid on the same homeowner, and they are judged by completely different rules.
Local Services Ads sit above the text ads, charge per lead rather than per click, and generally require background and license verification. Ranking there leans on reviews, responsiveness and how fast the phone gets answered. Eligibility and requirements vary by category and are worth confirming directly with Google.
Search ads give you the controls: keywords, negatives, ad copy, and a landing page you designed. That control is what lets you avoid the waste described above.
Run both if you qualify, but never merge the reporting. Cost per lead means something different in each product, and blending them produces a number nobody can act on.
Dispute the bad leads. Wrong trade, wrong area, spam. Left alone, they inflate the cost and quietly train you to distrust a channel that was working.
Price the click in windshield time, not in dollars
The only paid number worth a management meeting is what it costs to put an estimator in a driveway that turns into signed work.
Build the chain and measure every step: click, inquiry, contact, qualified, appointment booked, appointment held, quote delivered, job signed. Cost per lead alone rewards whichever campaign brings the most unbookable inquiries.
Push the outcomes back into the ad platform. Importing signed jobs and held appointments from your CRM as offline conversions lets the bidding optimize toward revenue instead of form fills.
Add the internal cost. An estimator crossing the Perimeter twice for a two-opening quote is a real expense even when the click was cheap, and it is the reason a campaign in a far county can look profitable on paper and lose money in practice.
Read the report by band and by product line. A blended metro number will hide a strong Cobb County result behind a Gwinnett one that never should have been running.
Questions we actually get
- How much should we budget to start?
- Work backward from capacity rather than from a published benchmark. Decide how many measures your estimators can hold in a week, estimate conservatively how many inquiries it takes to fill them, and fund the close drive-time band first. Adding budget before the intake and routing work is done just buys more inquiries nobody answers.
- Should we run Performance Max?
- Sometimes, and never as the whole account. It performs best with tight geographic targeting, brand terms excluded, aggressive placement exclusions, and conversion data that reflects signed jobs rather than form fills. Without those guardrails it tends to spend on cheap, low-intent traffic and take credit for demand you already had.
- Do Local Services Ads work for window and door replacement?
- They can, and eligibility depends on category and verification, which is worth confirming with Google directly. What decides the result is the phone. Ranking is influenced by review volume and responsiveness, so a shop that lets calls go to voicemail will pay for leads and lose them anyway.
- Should we bid on competitor brand names?
- It is legal to bid on the terms, though using another company's name in ad text generally invites a trademark complaint. Expect low quality scores and a poor close rate. Money usually works harder on the non-branded symptom searches unless a competitor is actively bidding on you.
- How do we tell whether the agency is just counting form fills?
- Ask them to report cost per held appointment and cost per signed job by drive-time band, and ask whether signed jobs are being imported back into the ad platform. If they can only produce leads and cost per lead, they are optimizing for a number that costs you estimator hours.
