A twenty five mile radius here buys you a lot of open water
Geography in this metro punishes circle targeting harder than it punishes almost anywhere else.
Draw a radius from a yard in Metairie or Mid-City and a large share of it lands on Lake Pontchartrain, the river and marsh. You pay for the impressions that fall on land, but the circle also reaches into places your crews will not go on a Tuesday.
Build the targeting from parishes and cities instead. Orleans as one group with its submarkets named. Jefferson with Metairie, Kenner and the West Bank. St. Tammany with Mandeville, Covington and Slidell held separately, because a job across the lake is a different day for a crew and should carry a different bid.
Set location targeting to people in your area rather than people showing interest in it. The French Quarter is full of visitors, and interest based targeting will happily sell you clicks from someone standing in a hotel lobby researching a contractor in another state.
Then look at the geographic report every month, not every quarter. Spend drifting toward places you do not serve is the most common leak in a local account and it is invisible from the campaign level.
In this market a window unit is an air conditioner, and it will eat the budget
The exclusion list is not housekeeping in this trade, it is the single highest return hour you will spend in the account.
Start with the words that mean something else entirely: window unit, window AC, tint, auto glass, plexiglass, stained glass, screen repair, glass repair, window cleaning and the operating system. Every one of them shows up in this trade's search terms and every one of them converts into a phone call that wastes ten minutes.
Add the intents you cannot serve profitably: one broken pane, apartment and rental repair, do it yourself measurement, sizes and dimensions, jobs and hiring, and anything shaped like a giveaway program.
Run exact and phrase match while you are building the data. Broad match belongs in an account that already has enough conversion history to steer it, not in a new one with a small monthly budget.
Read the actual search terms weekly for the first stretch, then monthly. Nobody enjoys it. It is still where most of the money is saved, and it cannot be delegated to an automated rule.
Custom sash and historic replica work is the expensive click worth paying for
The queries a franchise cannot fulfill are the ones where a local shop should be bidding hardest.
Wood windows, custom sizes, historic replacements, French doors, arched and non standard openings. Clicks on those terms cost more and convert slower, and they lead to jobs a volume dealer generally cannot quote because nothing about the opening is standard.
Group them apart from the general replacement terms so a bid strategy trained on quick conversions does not starve them. Mixed together, the cheap generic click always looks better in the report and the profitable one quietly loses funding.
Bid on your brands and on brand plus dealer phrases. Someone searching a manufacturer name and a submarket is deep in the process, and letting a lead aggregator sit above you on your own supplier's name is an expensive habit.
Keep entry doors and patio doors in their own campaign with their own budget. A door buyer is on a shorter timeline and a lower ticket, and blending them into whole house spend hides both.
Parade season empties the measure calendar, so pull the spend back
Pacing here is set by whether a crew can actually get to the house, not by the calendar month.
Streets close along the Uptown routes and downtown for stretches of the season, staging a truck becomes difficult, and homeowners postpone appointments they already booked. Buying full budget into a week when nobody will keep a measure is a straightforward loss.
Pace against install capacity in the other direction too. If the calendar is already booked well out, slowing spend beats paying premium prices for jobs that will sit and go cold while the customer collects two more quotes.
Heat is the demand driver in this market rather than a hard freeze. Bills climb, rooms stay uncomfortable, and older wood units get harder to operate. Plan for spend to rise ahead of that and not in reaction to it.
Write the pacing rules down and put them on a calendar with a named owner. Budget decisions made in the moment tend to follow whichever week had the loudest phone.
Ad copy that says raised house filters the callers your crews want
An ad is a qualifier as much as a hook, and the specific words keep the wrong caller from clicking.
Name the work. Wood sash, pier and beam access, openings that are no longer square, historic district experience, doors that swelled shut. A homeowner in a raised shotgun house reads that and knows you have been under their kind of house before.
Name the territory in the ad itself, by parish and submarket. It costs you a few characters and it prevents a slice of the clicks you would refuse anyway.
Use call assets with real hours, location assets tied to your profile, and a form on the landing page that is short enough to finish outdoors on a phone. Send whole house searches and single door searches to different pages, because a shared page always underserves one of them.
Say nothing about savings percentages, code compliance or approval outcomes in ad copy. Claims like that are hard to defend, easy to screenshot, and they generally set up a conversation your estimator has to walk back on the porch.
Import signed jobs into the account or Google optimizes toward the wrong caller
Left alone, the system will find you the cheapest form fill in the metro, which is rarely the job you want.
Track calls and forms, then send the outcome back. A booked measure and a signed contract are two different conversion actions with two different values, and the platform can only optimize toward what you feed it.
With the volume a single metro dealer produces, use booked measures as the optimization signal because there are enough of them to learn from, and judge the account on cost per signed job. Two numbers, two purposes.
Break the reporting out by parish. A lead from Kenner and a lead from Covington do not cost the same to serve, and averaging them across the account hides which side of the lake is actually paying for itself.
Give any structural change a full read before you judge it. Small monthly numbers move around on their own, and reacting to two slow weeks is how good campaigns get dismantled.
Questions we actually get
- What should we spend to start in this market?
- No honest number exists without knowing your average job value, close rate and service radius. The useful way to set it is backward: decide what a signed job is worth, estimate how many clicks and measures sit behind one, and fund enough weeks to gather data. A budget too small to produce steady booked measures cannot teach the account anything.
- Should the Northshore run in the same campaign as Orleans Parish?
- Keep it separate. A crossing to Mandeville, Covington or Slidell costs crew time that an Uptown appointment does not, which usually means a different bid and often a different close rate. Combined, the stronger market subsidizes the weaker one and you cannot see which is which.
- Is paid social worth running alongside search?
- It can be, but for a different job. Search catches people who already decided something must change. Paid social is where you reach the owner who has lived with a stuck door for three summers. Expect a longer path to a signature and do not judge it on the same cost per lead as search.
- Do Local Services Ads make sense for a window and door dealer?
- Availability by category and area changes, so it is worth checking directly rather than trusting a blog post. Where it is available it competes for the same phone as your search ads, so watch the total cost per booked measure across both rather than treating them as separate wins.
- How do we stop paying for repair calls we do not want?
- Exclusions plus copy. Build the negative list around repair, single pane, screens, tint and rental language, then write ads that state the scope you sell. Some repair intent will still get through, and a short qualifying question at intake costs less than trying to filter the last few percent in the auction.
