Google Ads for flooring companies: the setup that does not burn the budget

You turned it on. A month later you have a folder of calls: a landlord who wanted a patch kit, somebody asking whether you are hiring, two homeowners outside your radius, and one real estimate.

Nothing broke. Google did exactly what it was told. Nobody told it much, so it filled in the blanks, generously.

Google Ads answers two questions. What is a click worth to you, and what does a click you want look like. Most accounts answer the first and skip the second. The second decides whether the money buys measures or noise.

Search intent is the whole game

Two people search for flooring in the same hour. One is standing in a kitchen with a ruined floor and a demo crew booked. The other is on a couch deciding whether he likes wide plank, maybe next spring.

Both are flooring traffic. Only one is money this quarter. Every decision here comes down to telling them apart. Four kinds of searcher, wanting different things.

Read that as a budget instruction. Bucket one gets the money. Bucket two waits for the page. Bucket three waits longer. Bucket four gets blocked.

Campaign structure for a flooring company, and keep it simple

Most accounts are either one campaign holding everything or twelve holding nothing. Here is the rule that settles it.

A campaign holds a separate budget or a separate map. An ad group holds one tight idea and one landing page. If you cannot say what a campaign's separate budget is for, it should have been an ad group.

1. Brand. Your own company name. Small and capped, because a competitor can buy an ad above it. Nobody gets to show you brand numbers as proof the advertising works. Those people were already yours.
2. Service. One ad group per thing you sell and want more of. Hardwood install. LVP. Carpet. Tile. Bathroom remodeling and cabinets if you do them. One per job you want to quote, not one per product you stock.
3. Reach. Broad match, a second city you are testing, the price research bucket when you are ready. Its own budget, so the experiment cannot hide inside your best campaign.

Three to five campaigns, not twelve. Twelve campaigns splitting one modest budget is twelve campaigns starving, none of them with enough history to learn from.

One setting before anything else. Google's location targeting can include people merely interested in your area, not only people in it, and that is the more generous default. Set it to people in or regularly in your locations. Thirty seconds, and it is the most common reason a contractor fields estimates from other states.

Match types in 2026, and why broad needs a babysitter

Three match types. The practical difference is how far from your words Google is allowed to wander.

• Exact. Your term and things that mean the same thing. Tightest and priciest.
• Phrase. Your idea with words around it. The workhorse for service terms.
• Broad. The neighborhood of your idea. Cheapest clicks, widest reach, fastest way to fund a stranger's curiosity.

Broad is not the trap it used to be. It is fine when the negative list is maintained, the bidding points at a conversion that means something, and a human reads the search terms report every week. Take one away and it turns into an expensive random number generator.

The version that burns budgets looks respectable from outside. An agency turns on broad, points the bidding at form submitted, and never opens the search terms report. Six weeks later the account is excellent at buying form fills from people who wanted a price list or a job. Every chart is up. Your calendar is not.

So: money terms run exact and phrase in the service campaign, next to pages that match them. Broad lives in the reach campaign on its own budget. Never mix the two in one ad group, or you cannot tell which spent the money.

The negative keyword list, and where each one lives

Negatives are not housekeeping. They are the shape of the account. Keywords say what you want. Negatives say what you are, and in a trade where half the traffic is people buying boxes or looking for work, they do more of the job. They live at three levels, and mixing them up is how lists rot.

• Account level. Never want it, anywhere, ever. Build the list once and attach it to everything.
• Campaign level. Wrong for this campaign only. Brand terms go here, as negatives on the service campaign, so brand searches cannot get billed at service prices.
• Ad group cross negatives. The one nobody does. Hardwood terms excluded from the LVP ad group and back again, so each group serves only the ad and page that match.

A starter list for the account level, editorial, grouped by problem.

• Not the job you do. repair kit, patch, squeaky, buffing, cleaning, stain remover, plus commercial if you do only homes, refinishing if you do not refinish.
• Material without you. wholesale, supplier, distributor, remnants, per box, clearance, closeout, liquidation, sample, warehouse.
• Big box and marketplaces. Name every one near you. Somebody comparing retailers is not shopping an installer.
• Somebody who wants a paycheck. jobs, hiring, careers, salary, apprentice, course, training, certification, license.
• Doing it themselves. diy, how to, install yourself, tutorial, video, step by step, tool rental.
• Every town you will not drive to. By name, one at a time. Twenty boring minutes, routinely the biggest single saving in a new account.

Then read the search terms report weekly for two months. It will hand you five more nobody could have guessed from a website list, this one included. A longer version is in what flooring and remodeling marketing actually costs.

Local Services Ads and search ads are not the same purchase

Local Services Ads sit above the regular search ads, you pay per lead instead of per click, and getting in involves a screening step on license and insurance. With search ads you pay per click and control the message and the page. Coverage varies by trade and city, so go check your own category first.

Most run both. Local Services for the ready now calls, search for everything a category listing cannot say about you. One warning on the per lead model: disputing a lead that was clearly not yours takes somebody's time every week. Budget that time, or stop calling the per lead price the real price. Weigh a third source against the rest in the channel comparison.

The ad and the page have to say the same thing

Cheapest fix here, and the most often skipped, because it sits between two vendors and belongs to neither. If the ad says free in home measure this week, the first line of the page says free in home measure this week. Not "welcome to our website." The person clicked on a promise and spends three seconds checking whether they landed in the right place.

Four things above the fold, on a phone, and nothing else.

1. What you do, in the words the person typed.
2. Where you do it, by town name.
3. What happens next, in one sentence.
4. A tappable phone number and a short form. Both. The ones who call usually have a deadline.

One page per promise. If building pages is not realistic this month, point each ad group at your closest service page and rewrite its top screen. That still beats the homepage. And decide the form trade on purpose: fewer fields gets you more leads and worse ones.

Budget math, worked as an illustration

This is an illustration. The numbers are invented, round, and chosen because they divide cleanly. Not our results, not a projection for your city. Put your own numbers in the same slots.

One input is a range rather than an invention. A click on the highest intent flooring terms costs what the second most motivated bidder in your city will pay: a few dollars where three companies bid, well past fifteen in a crowded metro. Competition, not quality.

Now hold $1,250 against your own average job. Still in the illustration: if a job leaves two thousand dollars of gross profit, $1,250 to buy it is most of the profit and the plan does not work at these inputs. Which is a reason to know which lever you are pulling, not to quit.

Cheaper clicks, from tighter terms and negatives. A better click to inquiry rate, which is the page. A better inquiry to booked rate, which is the phone. Or a bigger average job, often the fastest of the four. What you are allowed to pay from your own margin is in cost per booked estimate.

Track booked estimates, not clicks

The click is what the platform counts for free. It is also the thing that has never paid for a van. Three layers, in this order.

1. Conversion tracking that fires on a real inquiry. Forms and phone calls, with calls over about sixty seconds counted separately from the eight second wrong numbers. If every ring counts, the bidding learns to buy rings.
2. Source tagging at entry. Tracking parameters on every ad link and a separate number per channel, landing on the contact record. If you cannot tell a search lead from a Local Services lead six weeks later, you cannot decide which to feed.
3. Booked and sold pushed back into the platform. Google accepts offline conversions. Once the bidding learns from estimates that got scheduled instead of from form fills, it starts finding people who answer their phone. Free, the strongest thing available here, and almost nobody turns it on.

Then the report. Five lines, weekly, same day. Spend, leads, cost per lead, booked estimates, cost per booked estimate. If the people running your search account cannot produce five lines, the problem is not the reporting.

The three ways owners get burned

1. Set it and forget it. The auction moves every week. A competitor raises a budget, a new company enters, the season turns. An account nobody has opened in a quarter is not stable, it is drifting toward the cheap clicks nobody wanted.
2. A dashboard that reports clicks. Impressions, clicks, click through rate, and a number labeled leads. Two questions end it. How many got a date on the calendar, and what did each date cost. A report that cannot answer both is decoration, and the reason is never technical.
3. One campaign for every service. Hardwood, LVP, carpet, tile and a bathroom remodel in one ad group, one ad, one page. The same generic ad goes to five people and four leave. Splitting them is the only way an ad can repeat back the words the person typed, which is the entire mechanism.

The shortest version

Buy intent, not traffic. Three campaigns, not twelve. Broad match only where somebody is watching it. Negatives before launch and every week after. Make the page finish the sentence the ad started. And judge all of it on what a booked estimate costs, because that is the first number in the chain nobody can improve without actually helping you. If the phone is the weak link, that fix is in the appointment setting playbook.