Meta Ads for real estate: fewer leads, better leads

A developer or brokerage will usually show me a dashboard with an enviable cost per lead, and then admit that the sales team has stopped calling the leads. Both facts follow from the same decision: the campaign optimises for the cheapest possible form completion, and Meta is extremely good at delivering exactly that.

The situation

Off-plan and ready-property campaigns running on instant lead forms, generating hundreds of leads a month at a very low cost each. Contact rate was poor. Of the people who did answer, most had no budget near the property price, many had not registered any real interest, and a meaningful share did not remember filling in anything at all — which is what happens when a pre-filled form sits one tap away from a video someone was half-watching.

The agency reporting and the sales floor reality had completely diverged, and nobody could point to which campaign had produced an actual viewing.

What the audit actually found

• Optimised for lead volume, not lead value. The campaign objective rewarded the cheapest form fill available anywhere in the audience.
• Zero form friction. Instant forms with three pre-filled fields, no qualifying questions, no review step.
• The offer attracted browsers. "Download the brochure" and "see prices" pull in people at the daydreaming stage, which is fine — as long as you don't then hand them to a closer.
• No CRM feedback loop. Meta never learned which leads turned into viewings, so it kept optimising toward the profile of people who fill in forms rather than people who buy.
• Broad, undifferentiated targeting. One audience covering everyone from first-time renters to investors, with the same creative.
• Pixel firing one generic event. No distinction between a brochure download, a qualified enquiry and a booked viewing.
• Slow follow-up. Leads arriving by email digest and being called the next day, in a market where response time is most of the outcome.

A cheap lead that nobody can sell to is not cheap. It costs the ad spend plus an hour of a salesperson's day, and it teaches the algorithm to find more people exactly like it.

The approach, in order

1. Define what a good lead is — with the sales team

   Before touching the ad account, I sit down with whoever answers the phone and get a written definition: budget band, purchase timeframe, cash or mortgage, resident or overseas, and which of those are deal-breakers. Everything after this depends on that definition existing, and marketing cannot write it alone.

2. Add friction on purpose

   Instant forms get switched from the fastest version to the higher-intent one: qualifying questions on budget and timeframe, a review step before submission, and copy in the form itself that repeats the price range. Volume falls immediately and that is the intended outcome. The people who complete a form after being told the price are a different population from the people who tapped through in two seconds.

   For higher-value inventory, I usually move off instant forms entirely and send traffic to a landing page. Slower, more expensive per lead, considerably better per viewing.

3. Rebuild the event structure

   Separate, meaningful events: a brochure download is not an enquiry, an enquiry is not a booked viewing. With the Conversions API sending qualified-lead and viewing-booked events back from the CRM, the campaign can finally optimise toward something that resembles revenue rather than form completions.

   This is the single change with the largest effect on lead quality, and it is the one that requires cooperation from whoever runs the CRM.

4. Then split audiences and creative

   Investors and end-users want different things and should never share an ad set. Investors respond to yield, payment plan and handover date; end-users respond to layout, location and lifestyle. I run separate ad sets with creative written for each, plus lookalikes built from the qualified-lead event rather than from all leads — a lookalike of unqualified leads simply industrialises the problem.

5. Then creative testing rounds

   Structured rounds, not endless variation: one variable per round, a fixed budget, a decision date. Price-led versus lifestyle-led hooks, agent-to-camera versus property walkthrough, single image versus carousel. Winners graduate into the evergreen set, losers get archived with a note about why.

6. Then fix speed to lead

   Real-time CRM integration, an automated first response within a minute, and a call attempt within five. In property this is not a nice-to-have — the difference in contact rate between a five-minute callback and a next-day callback is larger than most creative tests will ever deliver.

What changed

The pattern is always the same two-stage move. Week one to two: raw lead volume drops and cost per lead climbs, which looks like failure on the surface dashboard. Week four onward: contact rate improves because the people submitting forms now expect a call, and cost per booked viewing — the number that actually matters — starts falling below where it began.

The CRM feedback loop is the slow burner. It needs a few weeks of qualified-lead events flowing back before Meta's delivery visibly shifts, and it keeps improving for as long as the data keeps flowing.

What I would do differently

I would build the CRM integration in week one rather than week three. Every day the campaign runs without qualified-lead events being sent back is a day spent teaching the algorithm the wrong lesson, and that learning has to be unwound afterwards.

Common questions