Why Lead Ads Produce Bad Leads

Lead ads produce the cheapest cost per lead in the account by a wide margin. They also produce leads who do not remember filling anything in.

The platform is not underperforming. It is performing precisely as instructed, and the instruction was wrong.

You optimised for the form, not the client

Meta’s engineering team published on Andromeda, its ads retrieval engine, in December 2024. They reported a 6% recall improvement delivering an 8% ads quality improvement on selected segments, built on an increase in model complexity of roughly ten thousand times, and advertisers using Advantage+ AI features reporting a 22% increase in return on ad spend.

The relevant point is what that capability does when pointed at the wrong target. A system that good at finding people will find you an enormous number of people who complete a form, if completing a form is what you told it to value.

Form completion and buying intent are different behaviours. Optimising hard for the first actively selects for people who fill in forms easily, which is close to the opposite of what you want.

The format removes every point of friction

An in-platform lead form pre-fills the name, the email and the phone number, and submits in two taps without leaving the feed.

Every one of those conveniences removes a moment where somebody would have decided whether they actually wanted this. Nobody visited your site, nobody read anything, nobody saw a price. The threshold for submitting is a thumb movement.

That friction was not waste. It was qualification, and removing it is why the volume went up and the quality went down at the same time.

The pre-filled contact details are frequently stale

A practical problem underneath the strategic one.

The number pre-filled is whatever the person put in their profile, possibly years ago. A meaningful share are dead. Your team rings, gets nothing, and concludes the leads are fake. They are not fake, they are old, which is a different problem with a different fix.

What to change

• Turn on the higher intent form option. Every platform has one. It adds a confirmation step and cuts volume substantially, and the leads that survive it are a different population.

• Add a question only a real prospect can answer. Budget range, timeframe, property type, project size. One question, typed rather than selected, does more qualifying than any targeting change.

• Make them type the phone number. Removes the stale data problem and adds back a second of deliberation.

• Say the price on the form. A line stating that projects start at a number will cut submissions and raise every downstream figure. It is the single most effective change available here.

Then fix what you optimise toward

The deeper fix is to stop rewarding the platform for form fills.

Send a signal back when a lead becomes a real conversation, and again when it becomes a client. Then the delivery system is learning from outcomes rather than from submissions, and that ten thousand times model complexity starts working on the right problem.

This is the step most businesses skip because it requires connecting the ad platform to the CRM, and it is worth more than every targeting adjustment combined. A system optimising toward closed revenue behaves completely differently from one optimising toward a tap.

Where AI earns its place here

Two things, and the first solves the immediate problem.

Speed of response, at volume. Lead ad enquiries decay faster than any other source because there was no commitment behind them. Contacted within minutes, they convert. Contacted the next morning, most are gone. Nobody staffs for that, and an AI step that responds immediately with something specific rather than an autoresponder is what makes the channel viable at all.

Reading quality before the numbers exist. A system reading the actual submissions can tell you in week one that these are the wrong size of business in the wrong region, months before cost per acquisition would have shown it. That is the difference between wasting a quarter and adjusting in a fortnight.

It also does the closing of the loop described above, which is unglamorous data work nobody does by hand.

Judge the channel on the right number

The reason this persists is that lead ads win on the metric most accounts are managed by, and lose on every other one.

Compare channels on cost per closed client, not cost per lead. A source producing leads at $40 that close at one in eighty is more expensive than one producing them at $300 that close at one in five, and the second is also cheaper in sales time, which never appears in any advertising report.

Add the time cost explicitly. A hundred bad leads is roughly two days of somebody’s week, and if that person is on commission, you have also damaged their earnings and their willingness to work the list at all.

Ask the salespeople before you look at the data

They will tell you within thirty seconds which sources produce work and which waste their time, and they will be right.

The reason to check with them first is that a channel the team has quietly stopped calling is producing nothing regardless of what the platform reports. That is a fact you cannot see in any dashboard, and it means the fix is a conversation about trust as much as a change to a form.

When lead ads are genuinely right

They are not a bad format, and it is worth saying where they work.

Simple, low consideration offers where the next step is a phone call and the qualification happens there. Event registrations. Anything where volume genuinely is the objective and the cost of a wasted call is a minute.

They are wrong where the sale needs understanding, where a wasted call costs a salesperson twenty minutes, and where your team’s willingness to call the list is the actual constraint. A channel your sales team has stopped trusting produces nothing, whatever the cost per lead says.

Sources

• Meta Engineering, Andromeda: Advantage automation, next generation personalized ads retrieval engine, 2 December 2024.

AI Optimize answers lead ad enquiries within minutes with something specific, and sends closed revenue back to the platform so it optimises toward clients rather than taps. That work sits under Paid Ads Management.