Facebook ads for home remodeling companies

Start with who is in the room, and in this trade there are two of them. Pew's November 2025 survey puts Facebook at 80% of 30- to 49-year-olds and 74% of 50- to 64-year-olds, against Instagram's 62% and 40% — the bands that own the houses and hold the equity that pays for the work. But the number that decides the platform is not the reach of a person, it is the reach of a household. A kitchen is not bought by whoever saw the ad; it is bought after a conversation between two people who both have to agree, and Meta sells you one of them at a time. Facebook's advantage in those two bands is really an advantage in the odds that the second decision-maker is reachable at all — by you later, or by their partner tonight, with a screenshot.

That has a design consequence and it is the most underrated thing on this page: the ad has to survive being forwarded. Whatever the ad set knew about the reader — that they were in the 1950s ranch neighbourhood, that they had already visited the site, that they were shown the addition creative rather than the bathroom one — none of it travels with the screenshot. The version of the ad that arrives at the other owner has no targeting, no landing page and no context. Write for that arrival: name the business, name the offer, and put the number in the image where a crop cannot lose it. Every ad in this trade is really two ads, and the second one is the one that closes.

Third, the surfaces. Facebook has nine image placements against Instagram's four — Feed, in-stream video, video feeds, Marketplace, Stories, ads on Reels, search results, Business Explore and the right-hand column — and two of them matter here for reasons that are specific to remodeling. Marketplace is where a household already goes to buy a house's worth of objects, which makes it the one surface where a remodeling ad is not an interruption. And the right-hand column and search results have no Instagram equivalent at all: that is where the 1.91:1 export routes, and a wide export is the right shape for the documents this trade argues with. A schedule, a ledger or an allowance sheet loses nothing in a wide crop. A kitchen loses the ceiling and the floor.

Fourth, shape and the copy budget, measured on this batch the way every Facebook page here measures its own. Meta's ads guide recommends 4:5 at 1440 × 1800 for Facebook Feed, primary text of 50–150 characters and a 27-character headline, with no description line shown; Marketplace wants a square at 1080 × 1080 with 125 characters, a 40-character headline and a 30-character description. Leave Advantage+ placements on and one export fills all nine. This run wrote fifty headlines and the result is the best in the cluster: five fit 27 and all fifty fit 40 — the first batch here where every headline clears Instagram's limit. None of the ten lead headlines fits Facebook's own 27, which is the same result HVAC got. Eight of the ten description lines fit Marketplace's 30, in a field the feed will not show.

Fifth, the finding that changed on this batch, because seven pages on this site currently state the old version of it. Across the seven Facebook batches measured before this one, not one of 350 primary texts fitted the 50–150 recommendation and the shortest string ever written was 200 characters — stated as a floor rather than a tendency. Eleven of this batch's fifty-five fit, and the shortest is 64. But eight of these ten ads still produced nothing under 150, and all eleven short texts come from three ads, each of which wrote a standalone hook as its own variation rather than as the opening line of a body. The live-in kitchen ad did it five times out of five, at 74 to 99 characters, on the first pass. So length is not a property of the pipeline: it tracks how much of the argument the image is already making. Give the picture the whole case and the copy comes back short enough to fit.

Sixth, what Facebook is not for in this trade, because it explains why the Instagram case is a different business rather than the same ads reformatted. Remodeling's native visual is the finished room, and the honest problem with it is that every finished kitchen looks like every other finished kitchen — a slider between the dark oak and the island proves nothing a competitor's slider would not also prove. The version with information in it is a document: an allowance sheet, a completion date, a build calendar, a lien-waiver stack. Five of the ten briefs in this batch put paperwork in the frame for exactly that reason, and paperwork reads as evidence on this platform and as a bad photograph on the other one. This page is about the argument. The photogenic half of the trade is the twin.

Ads Manager offers six objectives — Awareness, Traffic, Engagement, Leads, App promotion and Sales — and a remodeling company has a real use for three. The batch on the examples page split six Leads, three Traffic and one Awareness, with no Sales ad in it at all, and that is the fourth trade in this cluster with no Sales line. The reason is ordinary: nobody buys an addition in a checkout.

Leads is the default and takes most of the budget, but the objective is the easy half of the decision and the performance goal underneath it is the hard one. Optimising for the lead is the standard mistake here and it is worse in this trade than in any other on this site: a large discretionary purchase attracts an enormous amount of browsing, and lead optimisation will faithfully find the people most willing to fill in a form about a kitchen they have no intention of buying. The destination section covers the mechanism and the budget section covers whether you can afford to use it.

Traffic is a strong second and unusually strong for a local service, because half of what a remodeler sells is a document the homeowner has never been shown — an allowance schedule, a payment schedule, a change-order rule, a lien waiver. Three of the ten ads on the examples page are Traffic buys and all three teach a piece of paperwork before asking for anything. The click is cheaper than the lead and, in a trade where the lead is worth qualifying, a page read is a better filter than a form.

Awareness is the third, and the argument for it here is neither the roofer's nor the heating contractor's. A roofer buys Awareness because a competitor will physically be at the door within seventy-two hours. An HVAC company buys it because demand is event-triggered and being remembered in January is the product. A remodeler buys it because the decision takes a year and is made in a conversation you are not in. What you are buying is being the name that is already in the room when two people finally sit down with a number — which is why the confession ad in the set below is bought as Awareness and judged on whether the phrase turns up later in sales conversations.

This is where a remodeling account is won or lost, and the losing version is easy to describe because almost everybody runs it: one before-and-after slider, pointed at everybody, with "free estimate" underneath. That ad competes with every other remodeler's identical ad, says nothing that transfers, and collects the cheapest possible lead — somebody who wanted a free thing. The offers below span a $250 assessment and a six-figure addition, and the person who has been eyeing one wall for four years is not weighing the same question as the couple deciding whether to sell instead.

Each row pins four decisions: who the offer is actually for, which angle family argues it best — every one is a page in the angle library — which objective buys it, and, where the examples page has one, a finished ad showing the whole thing assembled. Twelve rows is not a campaign plan. Pick the two or three that match what your crews are short of in the quarter that starts four months from now, because that is when the work these ads buy will actually be scheduled.

Two notes before you use the table. First, the cheap first transaction is the most underused offer in this trade and the one the account architecture wants. A remodeling ad's usual opening ask is a five-figure commitment dressed up as a free estimate; a fixed-price assessment at a named price is a real transaction, filters out the browsing, and — the part that matters — is an appointment that happens inside the attribution window, which is the only kind of event this account can optimise on. The structural-assessment row and the fixed-bid row are both versions of it.

Second, one row on this table carries a policy risk that the others do not, and it is the row with the best emotional argument in the trade. Aging-in-place work sells because somebody's parent is moving in, and Meta's personal attributes standard prohibits content that asserts or implies personal attributes, enumerates disability and physical or mental health, and extends to implying you know "the personal attributes of their family". Every natural sentence for that offer — safe for Mom, whenever she needs it — is the prohibited one. The compliant version sells the same job by describing the room instead, and the creative section has the fix.

Start from the declaration, because it is compulsory and getting it wrong is the most expensive mistake available in this account. Since 31 March 2020 every campaign created through Meta's Marketing API must specify a special ad category, including campaigns from businesses that offer none of housing, employment or credit, whose answer is simply NONE. Remodeling is not one of Meta's categories. So for a consultation, an assessment, a fixed-bid appointment or a build-calendar ad, a remodeling contractor keeps every lever: age, a radius below fifteen miles, ZIP codes, lookalikes, saved audiences, exclusions, all of it. The ordinary radius floor is 0.63 miles, or 1 kilometre.

Now the lever this trade lives on. A remodeler's audience is not a radius around the shop, it is a housing stock: the post-war capes with the galley kitchens, the 1970s split-levels with the sunken living room, the 1990s subdivisions whose builder kitchens are now thirty years old. That geography is ZIP- and neighbourhood-shaped, and unlike a roofer's hail swath it does not move — which is the good news and the reason the financed campaign hurts less here than there. The bad news is that most accounts never build it at all, and run a fifteen-mile circle that includes apartment blocks, new construction and three towns of houses nobody is remodelling.

Then the sentence this section turns on, and it is the same one every trade page in this cluster arrives at from its own direction: the category attaches to what an ad offers, not to what a company sells. Meta's scope is ads that "offer housing, employment, or credit opportunities". So a remodeling company is not in a category and a remodeling advertisement can be. Advertise a project on a fixed monthly payment and that campaign is a financial products and services campaign — the API value is FINANCIAL_PRODUCTS_SERVICES, required for US advertisers since January 2025, and any runbook still naming CREDIT has not been touched in over a year.

What the category costs, in full: location exclusion is unsupported; ZIP codes, neighbourhoods, sub-city, sub-neighbourhood, metro area, small geographic area and electoral district are all unsupported as location types; the radius floor becomes 15 miles or 25 kilometres in the US and Canada; age fixes to 18–65+ and gender goes to all; lookalike and saved audiences are removed; and detailed targeting loses behaviour and demographic targeting, interest exclusion and detailed-targeting exclusion, with surviving interests limited to a previously approved list. Custom Audiences survive in full, as do Advantage+ Audience and Detailed Targeting Expansion, and a list can be checked with the is_eligible_for_sac_campaigns flag before launch rather than at it.

Which produces this trade's version of a move every other page in this cluster has had to make, and it is the most permissive version of it. Real estate cannot describe the reader, because the housing category makes the copy the regulated surface. Apparel qualifies on the garment rather than the body. Remodeling is not in a category at all, so when the financed campaign takes the postcodes away, the copy may do the qualifying instead — and the thing it qualifies on is a building. "If your kitchen is original to a 1962 ranch" is a sentence about a house, and it will be read by exactly the households the ZIP list would have found. That freedom has one boundary, and the aging-in-place ad is standing on it: describe the room all you like, never the person in it.

One more thing before the table, because it fails as a hard error rather than as poor delivery. A business admin has to accept Meta's non-discrimination policy in Business Settings before a housing, employment or financial campaign will deliver at all — the API returns error 2859024, "Certification Required". Its neighbour 2909035 comes back for a custom age range, a saved audience, a lookalike or a radius under the floor. If the financing campaign will not launch, read those two codes before touching the creative.

The usual answer for a considered purchase is a landing page, and this is the one trade in the cluster where that answer is wrong for a mechanical reason rather than a conversion one.

Meta sells exactly one optimisation built for a funnel that finishes later, and it is called conversion leads: you send lead stages back from your CRM and delivery optimises for the stage rather than for the form fill. Its documented prerequisites are worth reading in full before you plan anything around it. The integration requires Facebook or Instagram Lead Ads — Instant Forms — with the 15–17 digit Meta Lead ID stored in your CRM; it asks that you generate at least 200 leads a month and upload at least once a day; the stage you optimise for must occur within 28 days of the lead; and its conversion rate must sit between 1% and 40%. Meta's own instruction on choosing the stage is to "select the earliest lead stage you would like to optimize for", because the system optimises for the down-funnel stages as well.

Read those requirements against a remodeling company and two things fall out. The signed build contract is excluded by the 28-day rule even for companies large enough to qualify, so the thing you actually sell can never be the optimisation target — which is the same conclusion the attribution window forced, reached again from the platform's own documentation. And what does fit inside 28 days, at a conversion rate in the low tens of percent, is the consultation that was booked and held. That is the event, and it is now the answer to two independent constraints rather than a preference.

The volume gate is the honest catch: 200 leads a month is beyond most single-crew companies, and a company that cannot feed the integration cannot buy the optimisation. The fallback is not a landing page, it is a harder form. Use an Instant Form on the "Higher intent" type, which adds a review step before submission — the "More volume" type is built to be filled from pre-filled profile data in one thumb movement, and a lead who never looked at what they submitted does not remember you when you ring back. Add three or four qualifying questions: what room, roughly what budget band, own or rent, and when they want it started. Accept the volume drop; in this trade the form is where the qualifying happens if the algorithm cannot do it.

Two things still get a landing page, and both for reasons other than conversion. The financing ad gets one as an obligation: the ad's whole promise is that the rate, the term and the total survive the click, and a page that reverts to "as low as" language undoes the ad and the disclosure it was carrying. And any ad whose proof is a document — the allowance sheet, the build calendar, the completion-date clause, the lien-waiver stack — gets a page that shows the document. An ad whose premise is paperwork should be able to produce the paperwork on the next screen. Never the homepage: somebody who tapped an ad about allowances and landed on a rotating hero of finished kitchens has been asked to start over.

Lead with the thing that is not about advertising at all, because it sits underneath the only event this account optimises on. A remodeling contract signed at the homeowner's kitchen table after an ad produced the appointment is a door-to-door sale. 16 CFR 429.0(a) defines one as a sale of consumer goods or services where the seller personally solicits the sale and the agreement is made away from the seller's place of business, at $25 or more when it happens at the buyer's residence — and the ad did the soliciting. Neither exclusion that looks relevant reaches it: the emergency exclusion needs a handwritten, signed, dated waiver describing a bona fide immediate personal emergency, and the home-repairs exclusion covers a visit the buyer requested for "repairing or performing maintenance upon the buyer's personal property", which a kitchen is not. So 429.1 applies: the contract carries the cancellation statement in bold face type of at least ten points and the buyer leaves with a completed Notice of Cancellation in duplicate. The consequence for this page is precise — the conversion you count on day three can be lawfully undone on day six, before the seven-day click window it was reported in has closed.

Second, money, and this trade has two separate rules to keep straight. Regulation Z, 12 CFR 1026.24(d)(1), makes the amount of any payment a triggering term: state one and the advertisement owes the reader the downpayment, the repayment terms over the full term, and the annual percentage rate using that phrase. Subsection (c) requires a stated rate to be given as an APR using that term, and subsection (a) limits an ad to terms actually available. The financing ad on the examples page does the disclosure correctly and then fails (a) on arithmetic: $60,000 over fifteen years at 9.49% APR is $626.17 a month and the ad says $609, four inches tall, in every one of its five primary texts. Never let a model produce an amortisation figure. Compute it from your lender's sheet, and paste it in.

Third, the deposit, which is the other money rule and it is state law rather than federal. California caps the downpayment on a home improvement contract: Bus. & Prof. Code § 7159.5(a)(3) says "the downpayment shall not exceed one thousand dollars ($1,000) or 10 percent of the contract amount, whichever amount is less", (a)(5) says the contractor "shall neither request nor accept payment that exceeds the value of the work performed or material delivered", and (b)(1) makes a violation a misdemeanour punishable by a fine of $100 to $5,000, up to a year in county jail, or both. "Half down to start" is a normal sentence in remodeling advertising and in California it advertises a misdemeanour. Other states cap it differently or not at all — check yours before a deposit term goes anywhere near an ad, and remember that a payment schedule is a term the ad can be held to.

Fourth, the personal attributes rule, which this trade meets in one specific place and meets hard. Meta's standard says ads "must not contain content that asserts or implies personal attributes", enumerates disability and "physical or mental health (including medical conditions)", and prohibits content implying the advertiser knows someone's attributes "or the personal attributes of their family". Aging-in-place work is sold in exactly those sentences — safe for Mom, whenever she needs it, she cared for you for decades — and every one of them is the prohibited version. Meta's own contrast is "Do you have diabetes?" against "New diabetes treatment available", and the fix here is identical: "curbless showers with blocking set for grab bars" sells the same job and asserts nothing about anybody. On the examples page that ad's violation is a caption burned across the foot of the image, where no headline swap can reach it, from a brief that mentioned no condition at all. Steering the brief is not the same as controlling the output; check the render.

Fifth, the two-decision-maker rule, which is a composition instruction rather than a policy. The ad that closes is usually the second time it is seen — forwarded, screenshotted, held up on a phone across a kitchen table — and none of the targeting travels with it. So put the identifying facts in the frame: the company name, the offer, the number the argument rests on. This is also the reason the wide 1.91:1 export is worth having for a document: it routes to Facebook's right-hand column and search results, and a schedule or a ledger crops to a wide slot without losing anything, where a room loses its ceiling and its floor.

Sixth, the pipeline's own failure modes, which this batch documented better than any before it. A visual direction that offers a slot for a statistic will have that slot filled — with a real number if the brief supplies one and an invented one if it does not, and this batch invented less than any previous run because five of its ten briefs put a document in the frame. Typography renders accurately; counted objects do not, and the failure has a new species here: the allowance board draws three bid cards under a caption reading "2 of them mean allowances", so the picture and the caption disagree about the picture. Rendered screens do not reconcile against their own chrome — an inbox with eight correctly descending timestamps sits under a status bar reading 8:14, so the top three arrived from the future. And the model will rewrite the advertiser to resolve a conflict rather than flag it: briefed for a remodeler, the stay-versus-move ad came back as a neutral analysis service and the financing ad came back as the lender.

Seventh, shape and length, measured on this batch. Facebook Feed wants 4:5 at 1440 × 1800, 50–150 characters of primary text and a 27-character headline, and shows no description line; Marketplace wants a square at 1080 × 1080, 125 characters, a 40-character headline and a 30-character description. With Advantage+ placements on, one export and one headline fill all nine surfaces. This run wrote fifty headlines: five fit 27 and all fifty fit 40 — the first batch on this site where every headline clears 40, and none of the ten lead headlines fits 27. Across the eight Facebook batches now measured, 28 of 400 headlines fit 27 and 379 fit 40.

And the finding that changed, which is the most useful thing this batch produced. Seven batches said the primary-text floor was absolute: 0 of 350 fitted the 50–150 recommendation and the shortest string ever written was 200 characters. This batch wrote eleven that fit, the shortest at 64 — and eight of its ten ads still produced nothing under 150. All eleven come from three ads, and in each the model wrote a standalone hook as its own variation rather than as the opening line of a body. The live-in kitchen ad did it five times out of five, at 74 to 99 characters, on the first pass. So length tracks how much of the argument the image is already making. If you want short copy, do not ask for short copy — brief a picture that makes the whole case and leave the copy nothing to explain.

Four stages, and the third one is unusual enough to be the reason this section exists: a remodeling funnel has a stage for a person who never saw the ad. The other three are ordinary, and keeping the whole structure this thin is deliberate — every extra ad set divides the same small pile of evidence, and in this trade the pile is smaller than in any other on this site.

The second stage is the one the account is judged on, and its performance goal is the held consultation rather than the lead wherever the volume allows. The fourth is a separate declared campaign rather than a stage, which is what the special ad category forces and what most accounts get wrong by trying to keep everything in one place.

The interesting one is the third. In every other trade here, retargeting means showing the ad again to the person who clicked it. In remodeling the decisive reader is often somebody else in the same house, who saw a screenshot rather than the ad, is not in any audience you own, and is the person actually holding the objection — the disruption, the money, the fear that it will not finish. So the warm stage carries the reassurance arguments rather than the offer, and it carries them in creative that reads correctly with no prior context. Site visitors and video viewers are the audience you can buy; the second decision-maker is the reader you write for.

Do the arithmetic before copying anybody's number. Meta's delivery system needs roughly 50 optimisation events per ad set per week to leave the learning phase; below that the ad set sits in "learning limited", which is not a penalty but is an admission that the system cannot optimise with what it is being given. Take your own cost per held consultation and multiply by 50. That is the weekly spend at which one ad set delivers with full information — and for a company doing twenty-five jobs a year it is usually a multiple of the entire marketing budget, because twenty-five jobs a year cannot produce fifty held appointments a week under any spend. The answer is fewer ad sets, not more, and the funnel above has four lines for that reason.

The conversion-leads gate is the second piece of arithmetic and it is a cliff rather than a slope. The integration wants at least 200 leads a month and a stage that lands within 28 days. Below that threshold the optimisation is not available at any price, so the money that would have bought lead quality has to buy it somewhere else: in the qualifying questions on the form, in the price on the entry offer, and in the geography. A paid assessment is a budget instrument as much as a creative one — it is the cheapest filter in this account and it is the only one that also produces revenue.

Then the thing that makes this trade different from every other one in this cluster, and it is a reporting problem that behaves like a budgeting problem. Spend today produces a consultation this week, a design agreement next month and a contract in the spring. So the account looks worst exactly when it is working hardest — the first ninety days show cost and appointments and no revenue, by construction, and the temptation to switch it off arrives before the first contract does. Decide the trial length in advance, in writing, and make it longer than your own sales cycle. An account judged on a thirty-day window in a hundred-and-twenty-day business will be turned off while the pipeline it built is still closing.

Two reporting habits make that survivable. Keep the offline upload — the Conversions API takes offline and store events with action_source set to physical_store within 62 days of the conversion, which reaches a design agreement for most companies and a build contract for very few — and treat it as your own bookkeeping rather than as something delivery learns from. And keep a spreadsheet the platform will never see, with the click date and the contract date on the same row, because the interval between them is the number that tells you how long to be patient, and it is the one number nobody else can give you.

Seasonality on top of all of it, offset by the lag. Kitchens finished in spring are contracts signed in autumn, so the autumn budget buys the spring calendar; January and February are the planning months and produce consultations rather than signatures; permit-heavy work — additions, structural, ADUs — wants a longer lead than anything else and should be advertised furthest ahead. The mistake is advertising the season you are in. Advertise the season you want to be building in.

The first four are all the same mistake wearing different clothes — treating a four-month sale as though it finished inside the reporting window. The rest are ordinary, expensive and common.