If you do not pull permits in West Lake Hills, stop paying for its clicks
The metro contains several incorporated cities with their own building departments, and an ad account that ignores that spends real money on work you will decline.
Google draws geography from city names and postal boundaries. Your business is drawn by which permit desks you are set up with and which crews you can send. The two maps do not match anywhere in this market.
Before touching bids, list the jurisdictions you actually work in. West Lake Hills, Bee Cave and Lakeway sit alongside city of Austin addresses and unincorporated Travis County, and the review path is not the same in each. Confirm what applies to you rather than assuming.
Set location targeting to people in your targeted locations, not people interested in them. Otherwise you buy the out of state investor researching a rental portfolio and the relocating buyer who has not closed yet.
Exclude aggressively at first, then add areas back one at a time as you confirm you can serve them profitably. An account that covers less ground and books more jobs is not a smaller account.
New construction in Leander and Kyle fills a window account with warranty calls
Fast-building suburbs generate a stream of searches that look exactly like your customer and are somebody else's obligation.
A homeowner in a four year old house with a failed unit is generally chasing a builder or a manufacturer warranty. They will fill in your form, take an hour of your phone time, and buy nothing.
Build the negative list around that reality: warranty, builder, new construction, sub, bid, wholesale, DIY, cost calculator, repair, glass only, screen, cleaning, and the search terms for services that are not yours.
Review search terms weekly for the first two months and monthly afterward. Broad match will find phrasings nobody predicted, and the drain shows up in the terms report long before it shows up in the pipeline.
Some repair intent is worth keeping if you sell service. Decide that deliberately, give it its own campaign and its own budget cap, and never let it share a budget with whole-house replacement.
Broad match and Performance Max will happily buy you the whole Hill Country
Automated campaign types drift outward and downward in intent unless you constrain them with data you control.
Performance Max is fed by your conversion signal. If that signal is a form fill, it will find you cheap form fills, which in this trade means small jobs and unqualified inquiries far outside your service area.
Start with exact and phrase match search campaigns on your core replacement and door terms. Add automation only after you have offline conversion data worth optimizing toward.
Keep brand searches separate so automated campaigns cannot claim credit for demand you already earned. Otherwise every report looks fine while new customer acquisition quietly stops.
Add account-level negatives and location exclusions to every campaign type, including the automated ones. Constraints are the only steering wheel you get.
Raise the budget in May, because August clicks land on a full install calendar
Paid demand for window work here rises with sustained heat, which is also when your capacity is tightest, so the pacing has to be planned rather than reactive.
Spend ahead of the season. Clicks bought while the calendar has room turn into measures that turn into contracts, and contracts booked in spring keep crews busy later.
During the peak, watch capacity rather than impression share. If quoted work is stacking up past what your installers can absorb, pull daily caps down instead of running the phone into the ground.
Ordered units mean the sale and the install are separated by weeks. Budget pacing should follow the sales calendar, not the install calendar, and the two are easy to confuse in a monthly report.
When you throttle, throttle geography and campaigns you can rebuild, not brand terms. Brand is the cheapest traffic you will ever buy and the first thing a competitor takes if you leave it uncovered.
Do not put a rebate promise in ad copy, because the service boundary decides that
Austin Energy is a municipally owned utility in a state where most customers pick from competing retail providers, so eligibility follows the address, not the city name in your headline.
Program terms change and boundaries do not follow the postal map. An ad that implies a rebate to everyone in the metro is a claim you cannot support at the kitchen table, and it costs you credibility at the exact moment you need it.
Write copy about what you can prove instead. Openings measured and ordered to size, what the crew does to the interior trim, who pulls the permit, how long a customer waits before anyone climbs on a ladder in general terms.
Use ad extensions for qualification: service areas, door and window as separate assets, and a callout that sets a realistic minimum job size if you have one.
If a program is worth mentioning, mention it on a landing page with a line telling the reader to confirm current terms with the utility serving their address. Vague honesty beats specific optimism.
Send signed contracts back into the auction or you are optimizing toward form fills
Cost per lead is a vanity number in a trade where a two-opening inquiry and a whole-house job arrive through the same form.
Set up offline conversion import so a booked measure and a signed contract flow back to the ad platform with their values. Bidding then moves toward the traffic that produced revenue rather than the traffic that produced typing.
At minimum, tag every inquiry with its source and follow it through your CRM to contract. Even a manual monthly reconciliation beats reading the platform's own conversion column as truth.
Report three numbers: cost per booked measure, cost per signed job, and average contract value by campaign. The third one usually reveals that your cheapest campaign sells your smallest jobs.
One signed whole-house job funds a great many clicks. Judge campaigns on that arithmetic, and give a campaign enough time to produce a handful of contracts before you kill it.
Questions we actually get
- Should we run search and paid social at the same time?
- Search first, until cost per signed job is stable and your intake handles the volume. Paid social can create demand for an upgrade a homeowner has postponed, but it is judged over a longer window and it needs a functioning follow-up process behind it. Adding it to a shop that cannot answer its phone reliably just buys more missed calls.
- What budget does an Austin window dealer need to learn anything?
- Enough that a campaign produces several booked measures a month, since decisions made on two or three data points are guesses. The right figure depends on your average contract value and close rate, both of which you should calculate before launching. Anyone quoting you a metro-wide cost per click without seeing your numbers is guessing too.
- Do Local Services Ads make sense for this trade?
- They can, and the vetting badge helps with buyers who moved here recently and cannot judge local reputation. Treat them as a separate channel with separate reporting, because they bill differently and attract a different inquiry mix. Watch whether they mainly deliver small repair calls before increasing commitment.
- How should we handle clicks from outside our service area?
- Exclude the areas outright rather than relying on the ad copy to deter people. Set targeting to physical presence, add explicit location exclusions, and check the geographic report monthly, because boundaries shift as platforms redraw them. Every excluded click is budget returned to the addresses your crews can reach.
- Our cost per lead went up after restructuring. Is that bad?
- Not necessarily. Cost per lead rises when you stop buying cheap unqualified inquiries, which is usually the goal. The number to watch is cost per signed job and average contract value alongside it. If both improved while cost per lead rose, the restructure worked.
