Davidson and Williamson deserve separate budgets, not one metro radius
A single radius drawn around downtown spends the same money on two buyers who convert at different rates and pay different average tickets.
Start with geography, not keywords. Build separate campaigns or separate budgets for Davidson, Williamson, Rutherford, Sumner and Wilson. Each gets its own bids, its own copy and its own reporting line.
The reason is drive time and job size. An estimator sent from Green Hills to Mount Juliet has spent a morning. If the average job in one county is smaller, the ad cost that county can absorb is smaller too, and a blended report will never show you that.
Use location targeting set to presence, not interest. Otherwise you pay for people researching from out of state, and in a metro with constant tourism traffic that leak is not small.
Bid Franklin and Brentwood knowing the buyer profile differs from East Nashville. Same product, different objection, different landing page. The account should reflect the difference rather than average it away.
Foggy glass, window cleaning and new construction will drain the budget first
The expensive clicks in this trade come from people who do not want what you sell, so the exclusion list matters more than the keyword list.
Build the negative list before the first campaign goes live, then keep building it weekly for the first two months. The search terms report is the only honest document in the account.
Common drains: cleaning services, blind and shade retailers, window film, DIY and how-to searches, glass repair for a single pane, screen work, new construction and builder supply, commercial storefront glass, and anything reading like a rental listing rather than a homeowner.
Repair intent is the hard one. A person with one failed seal is a real customer for somebody, and if you do not do that work the click is pure loss. If you do it, put it in its own low-bid campaign with a cheaper path so it never competes with whole-house budget.
Match types deserve discipline too. Broad match with a thin negative list is how a window dealer ends up paying for auto glass. Start tight, widen deliberately, and read the terms report before widening again.
The first hard freeze moves the market, and your daily caps should move with it
Demand for replacement windows here is weather driven, so a budget divided evenly across twelve months is wrong in both directions.
Cold snaps produce a surge in drafty window searches. So does the first big heating bill after a cold month. Pacing should anticipate those windows rather than discovering them afterward.
Pull budget forward into the periods when intent spikes and let the flat months run lean. A month split into equal daily caps will throttle you exactly when the auction is worth entering.
Watch the install calendar as the counterweight. Selling more work than you can schedule creates a long backlog, and a homeowner told the crew arrives in ten weeks will keep shopping. Pace against capacity, not just against the month.
Ice events change behavior for a week. Emergency board-up and repair queries rise, which is not your product. Consider adding those terms to negatives temporarily so the surge does not quietly consume the whole-house budget.
Ad copy that names the sash, the cold room and the ordered unit outperforms efficiency claims
Generic energy savings language is what every competitor writes, and it describes nothing the reader is currently living with.
Write to the failure. Windows that stick. Rooms that stay cold with the heat running. Condensation between the panes. Sashes that will not stay up. Specific discomfort earns the click that a stock efficiency claim does not.
Say what the process is. Free in-home measure, exact opening measurements, units ordered to size, a lead time you actually honor, one crew, cleanup. Process language filters out people who wanted a same-day repair.
Never promise a savings figure, a rebate or a code outcome. Say efficiency ratings vary by product and that eligibility for any incentive is worth confirming with the program administrator or your own accountant. Overpromising here creates disputes, and disputes become reviews.
Use assets to qualify. Sitelinks leading to insert versus full frame, an entry door page, a financing page and a service area page let the visitor sort themselves before they cost you an estimator's morning.
Meta targeting reaches the Hendersonville homeowner who has not searched yet
Search catches people who already decided to replace something, while paid social creates the decision in a market full of new arrivals.
Constant in-migration means a steady population of homeowners who have not started shopping and have no contractor to call. Search will not find them because they are not typing anything yet.
Use paid social for the before-and-after of a real Middle Tennessee house, a plain explanation of what drives price, and an offer that costs nothing to accept. The conversion is an in-home measure, not a download.
Keep audiences geographically honest. A campaign that spills across the whole state buys attention you cannot service. Build them county by county, the same way search is built.
Retarget the comparison readers. Anyone who read the insert versus full frame page is mid-decision. A follow-up answering the next objection is cheaper than buying the original click again.
One booked job pays for a lot of clicks, so measure to the contract and not the form
Cost per lead is the metric that makes a bad account look healthy, because it counts a fogged-pane inquiry and a twenty-opening whole-house buyer as the same event.
Instrument the whole path: click, form or call, qualified inquiry, booked measure, quote issued, signed contract. Push offline conversions back into the ad platform so bidding optimizes toward contracts rather than form fills.
Report cost per booked measure and cost per signed job, split by county. Those two numbers will usually disagree with the lead report, and the disagreement is where the money is.
Call tracking is not optional in this trade. A large share of window inquiries arrive by phone, often from a person standing in the cold room. Record and review the calls, because half of what looks like a media problem is an intake problem.
Give the account time before judging it. Sales cycles here run weeks because homeowners collect bids. Judging a campaign on a two-week window will make you kill the one that was working.
Questions we actually get
- What should we budget to start?
- Enough to hold consistent daily impressions in one county for a full month, then expand once cost per booked measure is known there. Spreading a small budget across five counties produces data too thin to act on. No honest agency can quote you a click cost for this market without running it, and any number offered before the account exists is a guess.
- Should we run Performance Max?
- With care, and never as the only campaign. It will find repair intent, brand traffic you already had, and search terms you cannot see clearly. Run standard search first so you know your real terms and your real cost per contract, keep a hard exclusion list, feed it offline conversion data, and only then test it alongside rather than instead.
- Is bidding on competitor names worth it?
- Sometimes, at a low bid, with copy that makes a genuine comparison rather than a claim about the other company. Expect a lower conversion rate and higher scrutiny. It works best when a competitor is well known in one county and you have real work to show there, and it is rarely where a first budget should go.
- Should we advertise in Murfreesboro if our crews are based in Davidson County?
- Only if the average job there covers the drive. Run it as its own campaign with its own budget and its own cost per signed job so you can see the answer within a season instead of guessing. Many dealers find the outer counties work at a smaller budget with a higher minimum job size attached.
- Most of our inquiries come by phone. How do we track that?
- Use dynamic call tracking numbers per campaign, record calls with the required notice, and log the outcome of every one in the CRM. Then push booked measures and signed contracts back into the ad platform as offline conversions. Without that loop the bidding optimizes toward whichever campaign produces the most cheap phone calls, which is usually the repair traffic you did not want.
