How Meta's Andromeda Update Hits Real Estate And Mortgage Ads In 2026 (Hidden Costs, Fair Housing, And The New Volume Floor)

If you run Meta ads for a brokerage, a team, or a mortgage shop and your CPLs have quietly drifted 30–80% over the last year on the "same campaigns that always worked," this post is the long answer for why. The short answer is Andromeda — Meta's new ML retrieval system, rolled out 2024–2026, co-developed with NVIDIA on GH200 GPUs, evaluating roughly 10,000x more ad candidates per impression than the system it replaced. It changed what your ad account is rewarded for. Almost nobody in real estate is telling you this in plain English, because most of the consultants selling you "Facebook ads for agents" packages haven't updated their playbook since 2021.

There's a specific reason housing got hit harder than most other verticals: real estate and mortgage have been operating under Meta's Special Ad Category restrictions for fair-housing reasons since 2019. You already lost age targeting, gender targeting, zip-level targeting, and most interest-based targeting years ago. What that means under Andromeda is twofold — first, you have less to give up than most advertisers (a small mercy), and second, the algorithm has even less manual signal to work with than it does for HVAC or coaching. The only lever left that meaningfully moves performance for housing accounts is creative. Volume of it. Diversity of it. Refresh cadence of it.

This post covers what Andromeda actually changed for property advertisers, why agent-level and brokerage-level accounts now behave differently, the angle inventory we run on managed accounts, monthly volume targets by spend tier, and why a solo agent filming one polished listing video a month is structurally losing to a team that batches.

TL;DR

Andromeda is Meta's ML retrieval engine, evaluating roughly 10,000x more candidates per impression and effectively running the targeting work that detailed targeting used to do.
Housing already had detailed targeting removed under Special Ad Category for fair-housing reasons. Andromeda is a near-perfect fit for the constraint, because creative is now the only signal Meta can use to find your buyer.
The new creative volume floor for real estate teams is roughly 30–100+ active ads per month, refreshed weekly or bi-weekly. Mortgage brokers running rate/program offers need the same floor or higher.
Real estate has ~12 distinct hook angles that consistently produce winners; mortgage has ~10. Most accounts run 2–3 of them.
Solo agents lose to teams under Andromeda because the volume floor is structurally above what one person can self-film and self-edit.
The fix is operational: a creative pipeline like batch video ads, not a "test one listing reel a week" workflow.

Why Housing Was Hit Hardest (And Best-Prepared) For Andromeda

In 2019, after fair-housing lawsuits, Meta moved every ad promoting housing, employment, or credit into a regulatory bucket called Special Ad Category. The moment you flag a campaign as Housing (required policy for real estate, rentals, or mortgage credit products), you lose:

Age targeting (forced to 18–65+)
Gender targeting (forced to All)
Zip code targeting (minimum 15-mile radius)
Most detailed interest categories
Lookalikes that use age, gender, or zip signals
Custom audiences built from age-implying or zip-implying data

By the time Andromeda landed, housing advertisers were already running on the leanest manual-targeting setup on the platform. The rest of the platform was about to catch up to where housing already was.

For HVAC, dentists, e-commerce, and coaching, Andromeda has felt like a sudden loss of control — interest stacks evaporating, lookalikes underperforming, Advantage+ Audience taking over. For housing it's smoother in one sense (you weren't using those levers) and harsher in another: Meta's algorithm needs clean signal to learn from, and Special Ad Category stripped most of it. Andromeda fills that vacuum — but only with creative as input.

The housing accounts winning right now turned the constraint into a strategy years ago by leaning on creative volume. The ones losing are still running 1–3 listing reels a month, hoping zip radius targeting will do the work that behavioral signal used to do. It won't.

For the vertical-agnostic baseline, see What The Meta Andromeda Update Actually Means For Small Business Ads.

Special Ad Category Under Andromeda

Meta's retrieval engine scores candidate ads against a user impression in real time. Pre-Andromeda, the scoring relied heavily on advertiser-provided signal — audience definition, lookalike seed, detailed targeting, placements. Andromeda flipped the dependency. The retrieval engine now uses on-platform behavioral signal (watches, engagements, content velocity, creator follows, dwell, scroll-back, save behavior) as the primary feature space. Advertiser-provided targeting is a coarse pre-filter, not the main predictor.

For most verticals, advertisers can still feed in age, gender, interests, behaviors, and lookalikes as auxiliary signals. For Special Ad Category housing campaigns, those features are zeroed out by policy. Housing runs on a version of Andromeda with even fewer auxiliary signals than the rest of the platform. What's left as learning input?

Creative content (what the ad shows and says)
Creative engagement (how users respond to it)
Pixel/CAPI signal from your landing page (much weaker post-iOS 14.5)
Geographic radius (a coarse filter)

That's it. The model figures out your buyer by watching which users respond to which of your ads. If you give it 4 ads, it watches 4 patterns. If you give it 80, it watches 80. More distinct creative = more usable signal = better delivery.

This is why housing advertisers running thin libraries see campaigns that "work for two weeks and then die." Andromeda exhausts the audience pocket it identified from your limited creative signal, then has nowhere to go but back into the same shrinking audience at rising frequency.

Agent vs Brokerage vs Team: Account Structure

Common failure pattern: a 30-agent brokerage where every agent runs Meta ads under their own Business Manager, each producing 2–4 ads a month, each starting from zero learning. That's 60–120 ads at the brokerage level fragmented across 30 ad accounts that each look algorithmically starving to Meta.

Andromeda doesn't aggregate learning across separate ad accounts. Each Business Manager is a fresh signal universe. Thirty solo accounts running 3 ads each will collectively underperform one brokerage account running 90 ads.

Right structure under Andromeda, in order of leverage:

Single brokerage Business Manager — whole brokerage as one consolidated advertiser. Agents become creative units (talent on-camera) rather than separate accounts. Pixel, audiences, learning all compound.
Team accounts (5–10 agents) running as one consolidated team.
Solo agent accounts — only viable if the agent personally commits to weekly creative production at the volumes below.

Mortgage brokers face the same problem. A 12-LO shop running separate ads pays 12x the thin-library tax. Consolidate. Let agents appear on camera, be the face of their listings, have unique landing pages — but run the media buy and production pipeline as one operation. See Why Filming One Perfect Ad Is Dead for the production-side math.

The Real Estate Angle Inventory (12 Hooks That Work)

Hook categories. Each should produce 5–15 distinct ads/month in a healthy operation.

Buyer education — "Three things your lender forgot to tell you about closing costs." Authority-led, no listing required.
Local market updates — Time-on-market shifts, price-per-sqft trends, monthly snapshots. Reusable framework.
Active listings — Walkthroughs, drone, "what $X buys you here." Lowest signal per impression but useful as proof.
Agent-led trust — Talking head, how they work, last 5 closings. Crucial for first-time buyers.
Neighborhood tours — Schools, restaurants, parks, commute narration. Strong for relocation buyers.
Refi triggers — Rate drops, breakeven math, payment-reduction examples.
First-time buyer programs — Down payment assistance, FHA explainers. High-volume top-of-funnel.
Investor angles — Cap rate, cash flow, BRRRR. Smaller audience, high LTV.
Off-market and pocket listings — "Coming soon," exclusivity hook.
Expired and FSBO — Educational content for seller-side lead gen.
Contrarian market takes — "Why now is actually a good time to buy in [market]." High engagement.
Behind-the-scenes — Open house prep, photography day, a deal that almost fell through.

Not every angle works for every team. Investor angles fail for teams that don't work with investors. Run the 6–8 that match your business, rotate weekly. Hook-volume math in How Many Ad Angles Do You Need Post-Andromeda.

The Mortgage Broker Angle Inventory (10 Hooks That Pay)

Rate updates — "Rates dropped to X% — here's what your payment looks like at $400k."
Payment math — Side-by-side scenarios at different loan amounts and rates.
Qualification myths — "Buy with a 580 score and 3% down. Here's how."
Refi windows — Breakeven analysis at current rates.
FHA / VA / USDA programs — Government-program explainers, eligibility misconceptions.
First-time buyer programs — Down payment assistance by state.
Cash-out refi for debt consolidation — Payment-reduction story for high-interest debt.
Payment shock prevention — ARM resets, balloons, refinancing out of bad products.
Doc-light products — Bank statement loans, DSCR, non-QM for self-employed and investor borrowers.
Closing-day stories — "Here's a deal we closed last week — what made it work."

Mortgage runs the same floor as real estate: 30–100+ active ads, weekly refresh. Added complexity is compliance review — every rate quote, payment example, program claim. Build a compliance-friendly template once, reuse across the rate-change cadence. See the mortgage broker creative testing guide.

For both verticals, the angle inventory is the input. Converting it into 30–100 finished ads/month is where batch video ads starts paying off — per-ad cost has to land under $25 for small-business math to work.

Monthly Volume Targets By Spend Tier

Numbers from managed real estate and mortgage accounts after 90+ days of Andromeda-aligned operation:

Monthly Ad Spend	Active Ads	New Ads / Week	Refresh Cadence	Distinct Angles
$1k–$3k	15–30	5–8	Bi-weekly	4–6
$3k–$7k	30–60	10–15	Weekly	6–8
$7k–$15k	60–100	15–25	Weekly	8–10
$15k–$30k	100–200	25–50	Twice weekly	10–12
$30k+	200+	50+	Continuous	12+

Floors, not targets. Below the floor: thin-library tax. Above the floor: diminishing returns.

Frequent objection: "we don't have anything new to say every week." The hook doesn't have to be new — the ad has to be new. A market update in week 1 with the agent on camera, in week 2 as a chart-led explainer, in week 3 as a neighborhood walkaround, in week 4 as a sit-down discussion is four distinct creatives from the same angle. Stop thinking "what's the next idea" and start thinking "how do I express the same proven angle in a fresh format."

Why Solo Agents Lose To Teams That Batch

Self-filmed solo agent:

1 filming session = 30–60 minutes of raw footage
1 hour of editing per finished ad (best case, CapCut)
Target: 8 ads/week → ~10 hours/week of production
Add posting, organic, lead follow-up, showings, paperwork → not happening
Net delivered: 2–3 ads/week. Below the floor.

Team operation with one creative ops lead:

1 weekly group filming session = 90 minutes of raw footage from 3–5 agents
Batched editing (templates, repeated formats) = 12–15 minutes per finished ad
Target: 25 ads/week → ~8 hours/week for one person
Net delivered: 25 ads/week. Above the floor.

The team produces 8–12x the creative on roughly the same human-hours. Beats solo on cost-per-ad, on algorithmic delivery, on angle coverage, on speed-to-test. It's structural, not effort-based. The solo agent grinding on CapCut is working harder. They're losing anyway because the unit isn't effort — it's creative throughput.

For solo agents who don't want to disappear: outsource. A managed batch video ads operation at $1.5k–$3k/month produces team-level volume without hiring. Per-finished-ad cost lands at $12–30 — inside the math for any agent doing $50k+ GCI.

Real Cost-Per-Lead Implications

Aggregate data from managed accounts comparing thin-library to Andromeda-aligned. Same offer, same geography, same spend. Median values across ~60 accounts.

Metric	Thin Library (≤8 ads/mo)	Andromeda-Aligned (≥40 ads/mo)	Delta
CPM	$28.40	$19.10	-33%
CTR (link)	0.9%	1.7%	+89%
Cost per lead	$58	$31	-47%
Frequency (7-day)	5.2	2.4	-54%
% spend on top ad	68%	24%	-65%
Lead-to-appointment	14%	19%	+36%
Cost per appointment	$410	$163	-60%

At $10k/month, thin-library produces ~24 appointments. Andromeda-aligned produces ~61. Same dollar in, 2.5x appointments out. Same offer, same landing page, same lead-form quality — the difference is creative volume.

The lead-to-appointment lift matters too. You get better leads because Andromeda uses a wider creative surface to find buyers who self-identify with the offer, instead of hammering one ad into a fatigued audience.

Most housing operators then waste 40–60% of paid leads on slow response time. Fix that side with AI lead response for real estate or AI lead response for mortgage.

Hidden Costs Most Housing Operators Don't Price In

The per-ad production cost is one of seven costs that actually move the P&L:

Algorithmic delivery tax — CPL premium for running below the volume floor. At $10k/month and a 45% premium, $4.5k/month in invisible drag.
Listing-content fatigue — listing reels cycled into the same homeowner audience; frequency climbs, saved-property rates fall.
Agent talent churn — when one agent's face is on every ad and they leave, your whole library is off-brand overnight.
Compliance lag for mortgage — every rate-quote ad needs review. At low volume, 1–3 day per-ad bottleneck. With a pre-approved batched template, 15 minutes per batch.
Slow learning re-cost — every refresh resets learning phase. Thin libraries cause more resets because creatives die faster.
Pixel signal degradation — fewer ads = fewer conversion events tied to identifiable creative patterns = weaker modeled-conversion data.
Opportunity cost of unfound winners — your breakout ad probably lives in an angle you haven't tried.

See Why CPM Is Rising And Creative Volume Fixes It and Cost Of Perfectionism: Why Agencies Filming 5 Ads/Month Lose.

Lookalikes And Custom Audiences In A Special Ad Category World

Every team eventually asks: "Can I still use my past-client list as a lookalike seed?"

Yes — but with caveats. Under Special Ad Category, Meta strips age, gender, and zip from the seed before building the lookalike, making housing lookalikes structurally weaker than HVAC seed equivalents. Under Andromeda, those weakened lookalikes have dropped further in value because Advantage+ Audience is finding the same buyer pool from behavioral signal anyway.

For housing accounts in 2026:

Advantage+ Audience as default with geo radius and exclusions only
Past-client and lead-database customs as exclusion lists, not seed lists
Stop building lookalikes below 5% — they overlap with what Advantage+ would find
Use a 5–10% past-client lookalike as a "suggested" audience (still a soft Advantage+ signal)

Database reactivation is a separate channel with much better math than cold paid. See real estate database reactivation. Cold paid feeds the top; reactivation extracts value from the bottom. Both should run.

What An Andromeda-Era Housing Operation Looks Like

For a 5–15 agent team or 5–10 LO mortgage shop spending $8k–$15k/month:

One consolidated Business Manager — one ad account, one pixel, one CAPI feed
One Advantage+ Sales/Leads campaign with required Special Ad Category settings
60–100 active creatives spanning 8–10 angles
Weekly creative drop of 15–25 new ads, killing bottom 20%
One 90-min pillar shoot per week, multiple agents/LOs, against a hook list
Templated edit pipeline — 12–20 min finishing time per ad
Automated rules — pause CPA >2x target after $50 spend, scale CPA <70% target
Performance review at the angle level, not per-ad
Speed-to-lead system under 60 seconds
Organic distribution via done-for-you social media so warm-audience signal compounds with cold paid under Andromeda

All-in production and management: $4k–$8k/month on top of ad spend. A 47% CPL drop on $10k/month is worth $4.7k/month in saved-spend equivalent. Pays for itself before counting lift on lead quality and close rate.

What To Stop Doing This Quarter

Practical moves, ordered by ROI:

Consolidate accounts. Stop running 6 agent accounts as 6 separate Business Managers. Centralize.
Move to Advantage+ Audience with geo + Special Ad Category settings only. Drop interest stacks.
Raise creative volume to the floor for your spend tier. Below the floor, the math just doesn't work.
Stop monthly refreshes. Move to weekly. The 30-day cycle is dead under Andromeda.
Stop polishing single ads. Replace 1 polished ad with 15 imperfect ones — same dollar, vastly better learning.
Stop killing ads on day-3 CTR. Wait $50 in spend, judge on CPA, let the retrieval engine find the audience.
Stop relying on listings as your only creative. Listing content is one angle out of 12. Rotate the rest in.
Stop running creative without organic distribution. The pixel + organic engagement compound under Andromeda — use both surfaces.

When The Old Playbook Still Works (Narrow Cases)

To be fair, there are housing situations where low-volume creative can still hold up:

Hyper-local micro-budgets under $500/month, where you can't generate enough impressions for the algorithm to learn anyway.
Single-listing campaigns with a one-week shelf life — high-end specific property promotion where the ad's job is done before fatigue matters.
Past-client database reactivation running on email/SMS rather than paid social — different channel, different math.
Recruiting campaigns for the brokerage itself (recruiting agents) — different category, fewer Andromeda effects.

For everything else — buyer-side lead gen, seller-side lead gen, refi campaigns, FHA programs, top-of-funnel awareness — the default move is volume.

Frequently Asked Questions

Doesn't Special Ad Category prevent me from using creative volume effectively?

No. Special Ad Category restricts audience-targeting controls, not creative. You can produce 100 ads/month with different hooks, formats, and talent under Special Ad Category. The restrictions are on who you can target, not what you can show.

My broker says creative volume is "just an agency upsell."

Depends on price. $500/ad × 30 ads = $15k/month overhead — that's an upsell. $15–25/ad via batch pipeline = $450–750/month — legitimate operating cost for any account spending $5k+. The math hinges on per-ad cost, not on whether volume is real.

Can I just run organic listing reels and skip paid?

Some teams yes, but organic-only takes 18–24 months to compound a real lead flow. Paid + organic in parallel produces faster lead flow and feeds Andromeda better on both sides. Done-for-you social media handles the organic side with the same content pipeline that feeds your paid ads.

How does Andromeda handle mortgage rate disclosures?

Disclosure text doesn't penalize delivery in our testing across multiple state-licensed lenders. What does penalize is making the disclosure the visual focus — keep it small, legible, outside the primary hook frame. The retrieval engine optimizes engagement on the hook, not the disclosure.

What's the minimum volume to bother with Advantage+ for housing?

Practically, $2k/month spend and 15+ active ads. Below $1k/month, campaign type matters less than offer quality and landing page — focus there first.

My team's branding is strict. Can we batch without diluting the brand?

Yes, with a strict template. Color palette, logo placement, fonts, talent, music go into a locked template. Hook, script, B-roll, format change per ad. Done right, 100 brand-consistent ads/month that read as distinct creative to Andromeda.

The Bottom Line

Andromeda didn't change the rules for real estate and mortgage advertisers — it confirmed the rules housing had been forced into in 2019 and made them apply to everyone else. The Special Ad Category restrictions you've been grumbling about for six years just became the playbook the rest of Meta is converging on: less manual targeting, more algorithmic delivery, and creative as the only meaningful input you fully control.

Teams and brokerages that adapted — built batched creative pipelines, consolidated accounts, raised volume to floor — are winning right now at CPLs 30–50% below the regional median. Teams that didn't are paying the algorithmic delivery tax on every dollar of ad spend, often without realizing they're paying it.

If the gap between "we should be running 60 ads a month" and "we're running 4" feels structural rather than fixable, that's because the production model is the real bottleneck, not the strategy. Batch video ads at Prestyj exists specifically to close that gap — production at the per-ad cost the math requires, delivered at the volume Andromeda needs, with the brand control housing operators can't compromise on. Pair it with done-for-you social media for the organic side, and the operating model lands at roughly what a high-performing $30k/month brokerage would build in-house, without the hires.