A Northern Virginia license does not stop Google from selling you Silver Spring clicks
Geography is the first setting to fix, because a radius drawn around a yard in this metro crosses two state lines before it stops being useful.
Set a radius around Alexandria and you will be bidding in the District and in Prince George's County within a few miles. Set one around Bethesda and you are buying clicks in Virginia. The metro is compact and the jurisdictions interlock.
Target by named locality rather than by radius. Build your geography from the list of places where you are licensed and where you will actually send a crew, and exclude the rest outright rather than leaving them at a reduced bid.
Also check the location setting itself. Presence rather than presence-or-interest keeps you from paying for someone in Denver researching a property they own in Dupont Circle. Some dealers do want those buyers, in which case run them as a deliberate, separate campaign, not as budget leakage.
Review the geographic report every month. Neighborhoods you never intended to serve show up quickly, and they show up as clicks before they show up as inquiries.
Bid Capitol Hill and a Fairfax subdivision at different numbers, because they are different jobs
Cost per click may look similar across the metro, but the job behind the click is not, and neither is what it is worth to you.
A whole-house replacement in a rowhouse means narrow access, alley staging, permit parking, custom sizes, likely masonry openings and possibly historic review. It is a longer job with more risk and, usually, a larger contract. A subdivision replacement with standard units and a driveway is faster and more predictable.
Split them geographically so you can bid what each is worth. Grouping the District, suburban Maryland and Northern Virginia into one campaign averages the value of every click and hides which side of the metro is actually paying for itself.
The split also lets you write different ads. Copy that mentions working in older masonry openings and staging from an alley reads as competence to a Shaw homeowner and as noise to someone in Reston.
Give each geography enough budget to gather data. Three campaigns starved of spend teach you nothing. If the budget cannot support the split, cut territory rather than diluting all of it.
Entry doors, patio sliders and whole-house replacement are three budgets pretending to be one
Product intent varies more than geography does, and mixing it in one campaign lets your cheapest job type eat the budget.
Someone searching for a front door replacement, someone searching for a sliding patio door that will not close, and someone pricing every window in a house are three buyers with different values and different close rates. Run them as separate campaigns with their own budgets.
Keep the landing pages matched. A whole-house ad pointing at a general services page loses people who wanted to see openings, materials and a sense of sequence.
Watch the door campaigns closely. Doors often draw a lower ticket and can convert well enough to look attractive on a lead report while producing less margin than the whole-house work you actually want.
If you sell installation of a specific manufacturer's line, that gets its own campaign too. Brand searchers are further along and cost less, and they should not be averaged in with generic terms.
Storm panels, sash cords and single broken panes will drain the budget by Thursday
Old housing stock generates a very large volume of repair searches, and none of them are your customer.
Broad and phrase match will pull in glass-only repair, screen repair, sash cord replacement, painted-shut windows, storm window fitting, weatherstripping and lock hardware. In a metro with this much century-old housing, that traffic is substantial.
Build the exclusion list from your own search terms report, not from a downloaded list. Run it weekly for the first month or two and then monthly. Every account we have seen for this trade has a different leakage pattern.
Also exclude the categories that share vocabulary and none of the intent: auto glass, commercial storefront glazing, window tinting, window cleaning, and blinds or treatments. They all buy the word window.
Consider what to do with repair intent rather than only blocking it. If you do offer restoration or storm panel work, give it a small deliberate campaign with its own economics. If you do not, the exclusion saves you real money.
The cold snap that spikes your clicks is the same week your crews are already booked
Demand here is seasonal and lumpy, and a flat monthly budget spends most in the weeks you can least use it.
The first hard freeze produces a rush of drafty-window and broken-glass searching. Competition rises with it and click costs follow. If your install calendar is already full into March, buying that spike aggressively is expensive attention.
Pace against capacity rather than against the calendar month. When the schedule is full, pull back on the highest-cost terms and keep a smaller presence on the searches that fill later slots. When the board opens up, push.
Shoulder seasons are usually the better buy. Late summer and early fall carry planning intent, less competitive pressure, and buyers who are willing to book a measure for a date some weeks out.
Dayparting matters more than most accounts use it. If nobody answers the phone after five, either cover the hours or stop bidding into them. An unanswered call on a paid click is the most expensive thing in the account.
A lead from the District and a lead from Fairfax County are not worth the same money
Cost per lead will mislead you here, because the same number hides three markets with different close rates and different contract sizes.
Report cost per booked measure and cost per signed contract, segmented by jurisdiction and by product line. Getting there means passing a lead identifier into whatever you use to track jobs and closing the loop when a contract is signed.
Without that loop, the account optimizes for the cheapest form fill, which in this metro is usually a repair inquiry or an out-of-area inquiry. It looks like improvement and it is not.
Call tracking is not optional for this trade. A large share of high-intent contact is a phone call from a phone, often from someone standing in the room. Record enough to know which campaign produced it and whether it was even in your territory.
Once the loop is closed, feed it back. Bidding toward confirmed booked measures rather than raw conversions changes what the platform chases, and it is the single change that most often improves the job board rather than the report.
Questions we actually get
- What should we budget to test paid search in this metro?
- Enough to gather data in one geography rather than a thin layer across all three jurisdictions. We would rather start with the jurisdiction where you have the most capacity and the best close rate, run it properly for a full season, and expand from proven numbers. A budget spread across the District, suburban Maryland and Northern Virginia at once usually produces three inconclusive campaigns.
- Is Local Services Ads worth it alongside search for window and door work?
- It is worth testing where it is available for your category, because the format sits above search results and is priced per lead. Screening and license verification requirements vary, and are worth confirming directly. Treat it as an additional channel with its own economics, not a replacement, and hold it to the same cost per booked job standard as everything else.
- Should we run paid social as well as search?
- Search catches people who have already decided something needs doing. Paid social can reach the household that moved here recently and has not decided yet, and it is reasonable for showing completed work in recognizable neighborhoods. It generally needs a longer read before you judge it, and it should be measured on assisted booked jobs rather than on cheap form fills.
- How do we stop paying for inquiries outside our licensed area?
- Three layers. Target by named locality and exclude everything else. State your covered jurisdictions in the ad copy and again above the form. Then ask for the property address as the first field, so an out-of-area inquiry is identified before anyone calls it. The third layer is where most of the savings actually come from.
- Can you guarantee a cost per lead?
- No, and anyone who does is guessing. Auction prices move with season, competition and how well your own site converts. What we will do is instrument the account so you can see cost per booked measure and cost per signed contract by jurisdiction, and make decisions from those instead of from a number nobody can promise.
