Why are Facebook lead ads producing unqualified leads?

Because the campaign is optimising for form submissions and the platform cannot see anything that happens after the form. With broad targeting, the conversion event is the only instruction it receives, so it finds the people cheapest to convert into a form fill, who are rarely the people most likely to buy. The fix is feeding qualification, site visit and booking data back through offline conversions, not adjusting the audience.

Do Meta instant forms produce worse leads than a landing page?

Generally yes, and the mechanism is friction. A pre-filled instant form asks the buyer to supply nothing and decide nothing, so it removes the free qualification that a landing page form with a phone field was doing. Higher-intent form types, a review step or two qualifying questions raise cost per lead and raise qualification rate; whether that is a good trade depends on which rose proportionally more.

Should I optimise my campaign for bookings instead of leads?

Not directly, at typical real estate volumes. Meta's published guidance has long put the learning phase at roughly 50 conversions per week per ad set, and a developer closing a handful of bookings a month cannot supply that. Optimise for the deepest event that still clears usable volume, usually qualified lead or site visit booked, and feed bookings back as offline conversions and conversion value so they shape the model without having to carry it.

What are offline conversions and why do they matter?

An offline conversion upload is a periodic file telling the ad platform what happened to leads after they left the platform: which were qualified, which booked a site visit, which turned up, which bought, and for how much. It is the only way the system learns a definition of quality that matches yours. Most accounts never implement it, which is why their lead quality never improves no matter how the targeting is adjusted.

Why are my Facebook and Instagram leads such poor quality?

The short answer

Because the system is working correctly against the wrong objective. Meta does not know which of your leads bought. It knows which of them submitted a form, because that is the only outcome you told it about. Ask for form fills and it will find the people most likely to fill a form and least likely to buy, since those are the cheapest people in the auction to reach. This is not a targeting failure to be fixed with better audiences. It is a feedback failure, and the fix is telling the platform what a good outcome actually was.

The optimisation event is the entire instruction

With broad targeting and Advantage+ placements, you are no longer choosing who sees the ad. The system is, and the only thing steering it is the conversion event you selected. Select Lead and you have instructed it, precisely and literally, to find people who submit lead forms cheaply.

It is very good at that. The complaint that leads are poor quality is usually a complaint that the machine did what it was told.

Why instant forms are the sharpest version of the problem

A native pre-filled form removes friction between the ad and the submission. That friction was not waste. It was doing qualification work for free, by asking the buyer to want it slightly more than a single tap.

Pre-filled instant forms produce the cheapest leads and the lowest intent, because the buyer supplied nothing and decided nothing.
A higher-intent form type, an added review step, or two qualifying questions each raise cost per lead and raise qualification rate. That is the same trade as disclosing approval status, and it is settled the same way: by whether qualification rose proportionally more than cost per lead.
Comparing your cost per lead against another builder's is close to meaningless when one of you is running instant forms and the other is running a landing page with a phone field. You are pricing different things.

What the platform never finds out

Everything after the form is invisible unless you send it. Meta has no view of any of this:

Event	Visible to the platform by default
Form submitted	Yes. This is the event you optimise on.
Telecaller reached them	No
Qualified on budget and intent	No
Site visit booked	No
Actually turned up	No
Booked a unit	No
Ticket value of that booking	No

So the model's entire concept of success stops at the form. Every optimisation decision it makes for the next month is built from a definition of a good outcome that your sales team would not recognise.

Three levels of sending the truth back

These are cumulative. Most accounts stop at the first, a good agency runs the second, and the third is the one that changes lead quality.

Browser pixel only. The baseline, and increasingly lossy: consent banners, tracking prevention and ad blockers all cut into it. Events that never arrive cannot teach anything.
Conversions API, server-side, deduplicated against the pixel. This makes the lead event reliable again. Worth doing, but note what it has achieved: it has made the platform more confident about the wrong objective.
Offline conversion upload. A weekly file saying which of those leads were qualified, which booked a site visit, which turned up and which bought. Only now does the platform have a definition of quality that matches yours.

Almost nobody runs the third one. It is the difference between an ad account that learns from your business and one that learns from your form.

What changes once bookings go back, and the trap in it

Two things become possible. You can optimise for a deeper event than the form, and you can send value with the conversion so the system learns the difference between a ₹45L buyer and a ₹1.5Cr buyer rather than treating both as one unit of success.

The trap: deeper events are rarer, and optimisation needs volume to learn from. Meta's published guidance has long put the figure at roughly 50 conversions per week per ad set to get out of the learning phase. If you close four bookings a month, pointing the campaign directly at bookings starves it, and performance gets worse rather than better while everyone concludes the whole approach failed.

The workable path is to optimise for the deepest event that still clears usable weekly volume, usually qualified lead or site visit booked, and to keep feeding the rarer booking events as offline conversions and as value. The rare events then shape the model without being asked to carry it.

The diagnostic, in order

Before blaming the creative or the audience, work down this list. In our experience the answer is above line four more often than below it.

What event is the campaign actually optimising for right now? Not what the brief said. Check the ad set.
Is the Conversions API live, and are server and browser events deduplicated? Double-counted events train the model on phantom success.
Are offline conversions being uploaded at all, on what cadence, and what is the match rate? A low match rate means the uploads exist and teach nothing.
Is lead source tagged at creation, in one system, so a booking can be traced back to the campaign that produced it?
Is the sales team's disposition data, the reason a lead was rejected, getting back to whoever runs the account? Otherwise nobody can tell a targeting problem from a qualification-script problem.
Only now: is the form type doing free qualification work, or removing it?
Only now: is the creative attracting the buyer or attracting the tap?

Questions people ask

Why are Facebook lead ads producing unqualified leads?: Because the campaign is optimising for form submissions and the platform cannot see anything that happens after the form. With broad targeting, the conversion event is the only instruction it receives, so it finds the people cheapest to convert into a form fill, who are rarely the people most likely to buy. The fix is feeding qualification, site visit and booking data back through offline conversions, not adjusting the audience.
Do Meta instant forms produce worse leads than a landing page?: Generally yes, and the mechanism is friction. A pre-filled instant form asks the buyer to supply nothing and decide nothing, so it removes the free qualification that a landing page form with a phone field was doing. Higher-intent form types, a review step or two qualifying questions raise cost per lead and raise qualification rate; whether that is a good trade depends on which rose proportionally more.
Should I optimise my campaign for bookings instead of leads?: Not directly, at typical real estate volumes. Meta's published guidance has long put the learning phase at roughly 50 conversions per week per ad set, and a developer closing a handful of bookings a month cannot supply that. Optimise for the deepest event that still clears usable volume, usually qualified lead or site visit booked, and feed bookings back as offline conversions and conversion value so they shape the model without having to carry it.
What are offline conversions and why do they matter?: An offline conversion upload is a periodic file telling the ad platform what happened to leads after they left the platform: which were qualified, which booked a site visit, which turned up, which bought, and for how much. It is the only way the system learns a definition of quality that matches yours. Most accounts never implement it, which is why their lead quality never improves no matter how the targeting is adjusted.