Facebook ads for real estate agents

The reach argument is the whole reason to be here and it is worth stating as arithmetic rather than as a vibe. After the special ad category has finished with your ad set, you have no age control, no gender control, no ZIP codes, no neighbourhoods, no lookalikes and no location exclusions. What you still choose is the app. Facebook reaches 74% of 50- to 64-year-olds and 57% of the over-65s; Instagram reaches 40% and 19%. If your offer is listing acquisition, buying Instagram is paying a reach premium to talk to your prospects' children. If your offer is a first-time-buyer explainer, the reverse is true and the other page covers it.

What Facebook gives back on shape, you should mostly decline. Instagram is portrait-only, which rules out the landscape listing photography every agent has a folder of. Facebook takes 1.91:1, which in our own publisher routes to the right-hand column and search results — placements Instagram has no equivalent for at all — while 4:5 goes to both feeds and 9:16 to stories and reels across both apps. The landscape slot is genuinely back. The trap is that the asset waiting to fill it is the wide-angle interior, which is the one thing every competitor in your market is already posting for free and which stops being relevant the week the house goes under contract. Both 1.91:1 ads on the examples page are text.

The copy budget is tighter here than on Instagram, which is the opposite of what most people expect. Meta's ads guide recommends a 27-character headline and 50–150 characters of primary text for Facebook Feed, against 40 and 125 for Instagram Feed. Twenty-seven characters is roughly four words. When we measured the batch built for this page, nine of the ten headlines the ads lead with were over it and 37 of the 40 alternates were too; all fifty primary texts exceeded 150, running from 261 to 563 characters. Longer is truncated rather than rejected, but at four words the truncation lands inside your first clause, so the rule is simply that the argument has to be intact before the cut.

And it is not one budget, which is what makes that rule binding rather than advisory. Facebook has nine image surfaces — Feed, in-stream video, video feeds, Marketplace, Stories, ads on Reels, search results, Business Explore and the right-hand column — against Instagram's four, and they do not agree. Marketplace recommends a square at 1080 × 1080, 125 characters of primary text, a 40-character headline and a 30-character description; Facebook Feed recommends 4:5 at 1440 × 1800, 50–150 characters, 27 characters of headline and shows no description at all. Leave Advantage+ placements on, which almost every campaign does, and one ad fills all nine from a single export and a single headline. You choose the file; you do not choose the frame. Eight of the ten ads on the examples page have a description short enough for Marketplace, in a field the feed will never render.

None of this changes because the campaign is declared housing — the special ad category takes targeting away and leaves the creative specifications exactly where they were. What it changes is the price of a truncated first clause. With no ZIP, no neighbourhood, no location exclusion and a fifteen-mile floor under the radius, the creative is doing the qualifying the ad set is no longer permitted to do, and the first forty characters are where it does it.

And one operational gate that stops a first housing campaign dead, which is not discoverable from the creative side at all. A business admin must review and accept Meta's non-discrimination policy before these ads will run: the Marketing API answers with error 2859024, “Certification Required”, and points at Business Settings. Its neighbour, 2909035, is what you get for a custom age range, a saved audience, a lookalike or a location radius under the floor — each with its own message naming the control. Do that at setup rather than discovering it on the morning a listing launches.

Ads Manager offers six — Awareness, Traffic, Engagement, Leads, App promotion and Sales. The seller side answers this differently from the buyer side in exactly one place, and it is the place the Instagram guide in this cluster says no: Awareness. That guide is right for buyers, where an unremembered brand is worth nothing because the decision is made in weeks. Sellers decide over years, inside a radius you already work, and being the name a household already knows when the conversation finally happens is a substantial part of what a listing agent sells. It is still the objective most often bought for the wrong reason.

Everything else is the same as it is anywhere in this category, and for the same reason: nothing about a house completes inside an ad session, so the most an ad can cause is a conversation. Seven of the ten ads on the examples page were written for Leads, two for Traffic and one for Awareness. None for Sales, and there is no version of this trade where that changes.

The rule underneath every row: the offer names something that happened to a house, not somebody it happened to. That is what makes it both a legal qualifier and a better one — a property event carries a decision, a demographic does not. Ten of the twelve link to a finished ad on the examples page; the last two have none on purpose, and say why rather than borrowing a neighbouring ad.

Note what is absent. There is no listing ad on this list. A boosted listing is relevant to the few people in a radius you pay for in full who want that specific house this month, and it expires at contract. Everything below survives the sale, which is what makes it worth putting a budget behind.

Everything the housing special ad category takes is documented on the Instagram guide in this cluster and is identical here — age fixed to 18–65+, gender to all, no lookalikes or saved audiences, no ZIPs, neighbourhoods or metro areas, no location exclusion, narrowed detailed targeting, a radius floor of 15 miles in the US and Canada and 15 kilometres in Europe. Read that page for the lever-by-lever version. What follows is what is different about running this on the seller side.

The first difference is that three of the most effective seller offers are also credit advertisements. A prep advance repaid at closing, a buy-before-you-sell programme underwritten against equity, anything rendering a mortgage pre-approval: those promote a credit opportunity, and financial products and services has been a required special ad category for US advertisers since January 2025 — Meta's own reference page gives the start date twice and disagrees with itself, 14 January in the banner and 21 January in the milestones, so treat it as live for anything you are running now. The good news is that its restrictions are the same ones housing already imposes, so declaring it costs you no additional reach. The bad news is that a wrong declaration is not a paperwork error — Meta's classifiers read the creative regardless, and being seen to work around a category is treated as evasion. If the ad advertises money moving before the sale, declare it.

The second difference is the certification gate, which catches almost every first-time housing advertiser. A business admin has to accept Meta's non-discrimination policy in Business Settings before any of this runs; until they do, the API returns error 2859024, “Certification Required”. It is not a creative problem, it is not a billing problem, and nothing in Ads Manager's campaign flow puts it in front of you at the moment you need it.

The third is the one thing on the seller side that genuinely works better than on the buyer side, and it is worth building deliberately. Custom audience inclusion survives the category. Your past-client list, your sphere, the people who opened your market report, everybody who watched the video about the tenant-occupied sale: those are addressable, and on the seller side they are the highest-value audience you will ever assemble, because a past client is a future seller with a known address. Meta's own documentation flags that customer lists carry an eligibility requirement and the same non-discrimination acceptance, so check before you upload.

The Instagram guide splits this by side of the transaction. This page is one side, so the split is different: it runs on how much the lead is worth and how fast you answer. A seller lead is the most valuable lead in this industry and among the rarest — one listing is a year of a small brokerage's marketing budget — which inverts the usual advice that more leads at a lower cost is better.

That is why the volume settings are wrong here. Meta's Instant Forms come in two types, More volume and Higher intent, the second adding a review step before submission. Everywhere else on this site, More volume is the sane default. On the seller side it fills your CRM with people who tapped a pre-filled form on the way past, and every one of them costs a phone call to disqualify. Use Higher intent, and ask one qualifying question that is about the property rather than the person — the address, or how long they have owned it.

Send the click to a landing page instead whenever the offer contains a number you have to defend: a published fee, a prep advance and its terms, a flat packet price. Those need the page before they need the form, because a seller who submits without understanding the terms is a seller who withdraws at the appointment. And where you have credit terms on the creative, the page is also where the disclosures live.

The third option is the one most listing agents under-use. Lead ads can click to message on Messenger, Instagram or WhatsApp, and the “Instant forms and Messenger” conversion location runs one campaign and lets delivery choose per person. For a seller thinking about it rather than deciding, a thread is a much lower commitment than a form and a much better conversation — but only if somebody answers within the hour. If nobody does, it is worse than a form, because an unanswered message is a bad first impression rather than an unworked lead.

Start with the rule the whole page turns on, because it is a writing instruction rather than a legal caveat. Qualify on the property, never on the household. “Your listing expired in March” and “you have a tenant with eleven months left” are precise, they select hard, and they are about a house. “Empty nesters”, “now that the kids have gone”, “perfect for a growing family” are about people, and familial status is one of the six classes the Fair Housing Act's advertising clause names. The test to apply to all five headlines every time: does this sentence describe something that happened, or does it describe who somebody is?

The second rule is about faces, and it is the one that surprises people. Eight of the ten ads on the examples page contain no people at all, and that is a defensible default rather than a stylistic one — in housing advertising the people you show are read as the people the housing is for, and HUD's 24 CFR Part 109, withdrawn in 1996 but still cited as HUD's position, addressed the selective use of human models directly. Where a face does appear on that page it belongs to the agent, which is the defensible use. If you are going to put people in a housing ad, do it deliberately.

Third: keep rendered documents short. This is a craft rule that came straight out of building the examples page and it has a sharp edge. Five of the ten ads render invoices, quotes or receipts, and the arithmetic held perfectly on the short ones — a four-line storage invoice, a three-line moving quote, a four-item prep bill all add up exactly. The one that ran to six itemised sections has three wrong subtotals and a grand total that does not match its own parts. Since the entire appeal of a document-realism ad is that a reader can check it, the length at which it stops being checkable is the length to stay under.

Fourth: twenty-seven characters. Facebook Feed's headline recommendation is shorter than Instagram's, so the argument has to be intact in about four words. On the examples page the headlines that survive that cut are the ones leading with a number and a noun — “12 homes. 7 taken. 5 left.” — and the ones that do not are the ones opening on a subordinate clause. Write the four-word version first and let the primary text carry the rest.

Fifth, and this is where a seller-side ad most often goes wrong on substance rather than policy: every figure you publish is a claim you will be asked to defend in a listing appointment. Days on market, list-to-sold ratios, repeat-client percentages, the price impact of a repair. The generator will invent all of them cheerfully and they will read beautifully. Replace each one with something from your own MLS or cut it — an ad that argues without a statistic is weaker than one with a real number and stronger than one with an invented number a client checks.

Finally, the rule that changed recently and that most competing guides still get backwards. Since HUD's Dear Colleague letter of 24 April 2026, sharing neighbourhood crime and school-quality information is permitted rather than prohibited, provided it is shared consistently and without discriminatory intent. That is a genuine expansion of what a seller-side ad may discuss. What did not change is the distinction underneath: objective data shared with everyone is information, and racially coded commentary about schools or crime remains evidence of intent even where the underlying data is accurate. Keep the source, drop the adjective.

The seller funnel is a calendar rather than a ladder, and that is the difference from every other page in this cluster. A buyer decides in weeks and a seller decides over years, so the job is not to accelerate somebody through three stages this month — it is to be present, credible and specific at an unknown future moment, and to capture the people who happen to be at that moment now.

Which changes what the middle stage is for. Cold is a circle you cannot shape, so its only job is to find reactions. Warm is where the productised first steps live and where most of the budget should go, because somebody who read the fee breakdown or downloaded the carrying-cost worksheet has raised their hand about a property rather than about a house. Sphere is the stage this platform makes uniquely valuable: custom audiences survive the special ad category, past clients are future sellers at known addresses, and on Facebook they are actually reachable.

Do the learning-phase arithmetic before you plan anything else, because on the seller side it comes out worse than anywhere on this site and pretending otherwise wastes the first two months. Meta's delivery system wants roughly 50 optimisation events per ad set per week to leave the learning phase. A small brokerage does not get 50 seller leads a week; it may not get 50 in a quarter. So you will be running learning-limited more or less permanently, and the correct response is not more budget on the same event — it is optimising for something further up that actually fires often enough to teach delivery anything, and judging the account on the rare event yourself.

That is also the honest argument against the home-valuation offer that dominates this industry. It is the one seller offer that produces enough volume to satisfy the learning phase, which is exactly why it is everywhere — and its leads are people who were curious two years before they will list. If you run it, run it knowing you have bought a follow-up problem rather than a listing, and staff for that. If you cannot work a two-year-old lead, the volume is a cost.

The calendar is the other half of the plan, and it is more specific in this trade than in most. The listing supply decision happens in the weeks after New Year, for a market that peaks in spring, which means the money should go in before the auction gets expensive rather than during it — the January intake ad on the examples page is that idea as a creative. The same logic applies in reverse in the autumn, when sellers who failed to sell over the summer are deciding whether to relist or withdraw, and expired-listing offers are cheaper and more welcome than at any other point in the year.

Judge the account on listing appointments taken and listings signed, not on leads. Cost per lead is the number Ads Manager shows you and it is the number that will mislead you fastest here, because the cheapest seller leads are systematically the worst ones. Two expensive leads that became appointments beat forty cheap ones that became a call list. No benchmark figures appear on this page for the same reason they appear nowhere else in this cluster: the ranges published online are averages across markets that have nothing to do with yours, and quoting one would be inventing a number in a guide that spends four paragraphs telling you not to.

The category mistakes — planning around ZIP codes, building a lookalike from past clients, hoping you will not have to declare — are covered on the Instagram guide and are identical here. These are the ones that belong to the seller side and to this platform, and three of them are visible in the finished ads on the examples page rather than hypothetical.