There are two versions of this story and most flooring owners have lived one. You boosted a job photo. It got likes from your cousin, a guy who sells siding, and one message asking whether you do repairs on a mobile home two hours away. So Meta does not work for flooring.
Or you ran real ads. Say forty inquiries in three weeks, invented numbers, at a price that looked great on the report. You called maybe nine of them, most did not pick up, you booked two. So Meta leads are junk.
Same problem, different clothes, and neither one is about Meta. Here is the order it goes in: what these ads are for, what the creative has to be, what targeting still does in 2026, where the inquiry lands, and the first five minutes after somebody taps.
What Meta is for, and what search is for
These two channels do different jobs, and that difference sets everything downstream.
On search, somebody typed "hardwood floor installers near me." They already decided. You are bidding on a buyer who is halfway out the door and comparing three companies. That is capture.
On Meta, nobody was looking for you. They were looking at their nephew's graduation photos. Your ad is an interruption. What it does that search cannot is make somebody who was not shopping decide that the kitchen floor they have ignored for four years is finally the thing. That is demand creation, and it is most of this market. Almost none of those homeowners will ever type a word into a search bar.
| Search | Meta | |
|---|---|---|
| The person | Already decided, comparing companies | Was not shopping, decides because of your ad |
| What wins it | Being on the right terms, answering first | The creative. Everything else is downstream |
| The inquiry | Higher intent, higher price | Lower intent, lower price, softer |
| How it fails | Budget too small to show up | Creative is stock photos and adjectives |
| What you owe it | Answer fast | Answer fast, then keep going for weeks |
That last row is the whole article. Meta hands you people earlier in their thinking than search does. Earlier means cheaper, and softer, and softer means more work between the tap and the truck. Treat one like a search lead and you will call it junk. It is not junk. It is early. For where this channel sits against everywhere else a flooring lead comes from, see the channels that actually produce flooring leads.
The creative is the targeting now
Fix one thing on this page, fix this. The old lever was the audience panel. The lever now is the ad itself. The system watches who stops and who fills the form, then goes looking for more people like them. Change the ad and you change who it hunts for. The ad is the targeting, which is why two companies with identical settings get results that are not in the same universe.
What earns a stop in this trade:
- Real rooms in real houses. Ordinary light, a phone held by somebody on the crew, furniture shoved against the wall. A showroom shot reads as an ad in the first half second.
- Before, then after, fast. Open on the worst of the old floor, the curling vinyl at the fridge, the gouged laminate by the back door. Then the same angle, done. The cut between them is the ad.
- The floor going in. Process footage proves a crew exists in a way a finished photo never does. The cuts, the underlayment, the transition strip. People watch work.
- The owner on camera. Your name, your town, one sentence about what you do that the cheap guy does not. Slightly awkward beats polished, because polished looks bought.
- Financing, said plainly. For a lot of homeowners the floor is a cash flow problem, not a want problem. If you offer payments, put the terms in the ad. Do not hint.
What reliably does nothing: stock photography, your logo on a colored square, the words quality craftsmanship, drone footage of your truck.
What to shoot on the next job
One job filmed properly is a month of ads, which is the whole idea behind creative cut off your own jobs.
- The walk in, before anything moves. Ten slow seconds of the old floor at its worst. Get low. Damage reads better from ankle height.
- Three or four cuts of the work happening. Tear out, subfloor, first plank, the transition. Ten seconds each is plenty.
- The reveal, from the exact spot as the first clip. Same corner, same height. The match is what makes it land.
- The owner and the homeowner, twenty seconds each. Say the town out loud. Ask the homeowner what made them finally do it. Their answer is your next ad.
Shoot vertical. Assume the sound is off, so anything that matters gets burned into the picture as text. And spend your effort on the first two seconds, because that is where the decision gets made.
Targeting in 2026, honestly
The old playbook stacked interests. Homeowners, plus recently moved, plus an income band, plus an age range, narrowed until the audience number looked satisfyingly precise. It felt like control. Mostly it bought a pricier slice of the same people.
After the tracking changes of the last few years the platform sees less of what happens off its own apps, so it leans on what it can still see, which is behavior on the ad. That is the honest reason broad plus strong creative wins now.
- Start broad. Location, an age floor that matches people who own homes, and little else. Let the creative do the filtering.
- Set location by drive time, not city name. Draw what your crew will actually drive, then exclude the towns you will not go to. That setting saves more than any audience you could build.
- Keep one narrow audience, which is retargeting. People who watched most of a video or abandoned the form. That is real, because they did something.
- Upload your past customers. Exclude them from the estimate ads, then run them a separate ad about the next room.
Do not trust the homeowner filters to keep renters out either. If renters are filling your form, the fix is the ad, not the audience panel. Judge targeting by what lands in your CRM, never by how impressive the reach estimate looked.
Where the inquiry should land
Two real options, and the answer depends on your office rather than on the ad.
| The form on the platform | Your own landing page | |
|---|---|---|
| Effort for the homeowner | Lowest. Prefilled, two taps | Higher. A page loads, then they type |
| Volume and intent | More inquiries, softer. Some taps are accidental | Fewer inquiries, firmer. They chose to type |
| What you can ask | A little. Every question costs volume | Anything, photos included |
| The hidden cost | Sits on the platform unless wired to your CRM | Somebody has to build it and keep it fast |
| Who it suits | A desk that calls within minutes | A team that cannot call fast |
Pick with one question: can somebody, in your office or on a team calling on your behalf, make first contact in five minutes, every time, including Saturday morning? If yes, use the form on the platform, because volume plus speed is the strongest combination in this trade. If no, use a landing page, because the friction filters and a slower follow up survives on a warmer inquiry.
Two rules either way. Ask exactly one qualifying question, which room or roughly when. And wire the inquiry straight into your CRM, never into an email notification, which is how a lead ends up read at midnight.
The follow up is the campaign
Most articles give the follow up one line at the end. It decides whether everything above was worth doing.
A Meta lead is somebody who was not thinking about floors ninety seconds ago. The interest is real and it is thin, with a half life measured in minutes. Two hundred posts later the memory of your ad is gone, and your call that evening comes from a stranger with an unknown number.
Speed is not a nice thing to have on this channel. Speed is the channel.
- Inside five minutes, text, do not call. Their name, the ad, one question. Hi Dana, it is Mike at the flooring company, you asked about vinyl plank, which room? A text gets read when an unknown number gets declined.
- Call about two minutes after the text. Now the ringing phone has context sitting on the screen. Different call entirely.
- No answer, try again the same day at a different hour, then keep going for several days across text, call and one voicemail. Somebody who cannot talk at four is often free at seven. One attempt and quit is where the leads you already paid for go to die.
- Then slow down instead of stopping. A check in every couple of weeks. A lot of these people were a someday, and whoever texted last gets the call when someday arrives.
- Every reply reaches a human in business hours. An automation that texts and then leaves somebody talking to nobody is worse than none.
Here is how these leads actually die, and nobody in the story does anything wrong. The inquiry lands at 7:12 on a Thursday evening as an email notification. The owner sees it Friday around noon, means to call after the last estimate, and does not. Saturday it is two days old and the call feels awkward. By Monday it belongs to somebody else. More on why that happens is in three reasons your leads ghost you, and what running that contact layer with a real person looks like is in we call your leads.
Budget, worked as an illustration
An illustration, using round invented numbers. Not our results and not a projection. It is arithmetic, so put your own figures in the same slots.
- You want 6 jobs a month out of this channel.
- You close 1 in 4 of the estimates you attend from it, so you need 24 attended.
- About three quarters of booked appointments get attended, so you need 32 booked.
- About 1 in 4 inquiries books an appointment, so you need 128 inquiries.
- At $35 an inquiry that is $4,480 of ad spend. At $65 it is $8,320.
Two things here are ranges rather than illustrations, with the reasoning attached. An inquiry on this channel usually costs less than the same one on search, because you are buying attention instead of intent. And fewer of them turn into booked appointments, for the same reason. Those move in opposite directions, which is why comparing channels on cost per lead tells you nothing.
There is also a mechanical floor under a budget, and it is not a sales line. The platform's own long published guidance puts the exit from the learning phase at roughly fifty conversions per ad set per week, which at any realistic price is not a number a contractor hits. So a sensible operator points the campaign at a cheaper event further up the chain and accepts slower learning. What follows is simple: do not split a small budget across four ad sets, do not kill an ad after three days, do not change two things at once.
How to tell if it is working
Not cost per lead. Never cost per lead by itself. It is the number a vendor can cut in half by making the form easier while your calendar stays exactly as empty.
Six lines, every week:
- Ad spend, with fees on their own line.
- Inquiries delivered.
- Inquiries contacted, and the median minutes to first contact.
- Estimates booked.
- Estimates attended.
- Jobs sold, and total sold value.
Divide the spend by the attended estimates and you have the only number worth arguing about, which is laid out in full in cost per booked estimate.
Two channel specific things sit alongside it. How many people are still watching a few seconds into your video, because if almost nobody is, the problem is your opening and not your budget. And how often the same people keep seeing the same ad, because when that climbs and your cost climbs with it, the answer is a new creative. That treadmill is the real reason to film every job.
The shortest version
Meta finds people before they start looking, which is most of your market. The creative decides who sees it, so film real jobs and stop buying stock. Go broad and let the ad do the sorting. Put the form wherever your desk can answer it fastest. Then treat the first five minutes as the product, because a lead nobody called is the most expensive thing in your business.
To watch the chain move when you change one input, the calculator on the homepage runs the same arithmetic here: run your numbers.
