The form that fills itself

There is one advertising format that fills in the form for the customer. They tap the ad, a form rises up the screen already carrying their name, their email address and their phone number, and two taps later you have a lead. No page loads. Nothing is typed. Nobody is asked to leave the app they were already in.

It works exactly as designed. Cost per lead on Meta instant forms is routinely a fraction of what the same business pays for a form fill on its own website, and the volume arrives immediately. Then the sales team starts calling, and the story from the phones does not match the story in Ads Manager: numbers that ring out, numbers that belong to somebody else, people who do not remember filling anything in. The account looks like a triumph and the pipeline does not.

Both of those observations are correct, and the reason they can both be correct is sitting in the mechanism. The form filled itself. The contact details in it came from a profile rather than from a person deciding to hand them over, and that difference shows up everywhere downstream — in answer rates, in match rates, and eventually in the revenue report, where the channel that produced the most leads produces the least traceable income.

A prefilled form measures how easy the form was to submit. Every other channel's form measures, at least a little, how much the customer wanted to submit it.

What an instant form actually collects

A Meta instant form is prefilled from the person's Facebook or Instagram profile: the contact fields, the name and, where the advertiser asks for them, work and demographic details are copied in from what Meta already holds, and the person can edit them before submitting.

This is not a hidden mechanism. Meta documents the prefill fields precisely, and the list is longer than most advertisers realise: email, phone number, WhatsApp number and full postal address; first and last name; date of birth, gender and marital status; job title, work phone, work email and company name. There is a vehicle block for eligible advertisers, carrying VIN and licence plate, and a national ID block for several Latin American markets.

The advertiser chooses which of those to ask for, and each one they choose is one the customer does not have to type. That is the whole product. It is also the reason the leads behave the way they do.

Read the second of those notes again, because it inverts the usual assumption. The prefilled value is not a verified value. It is the value on the profile. If the profile has no phone number, the person is asked to type one; if it has one from four years and two handsets ago, that is the one that arrives in your file, and the person is under no obligation to notice.

A prefilled number is not the same object as a typed one

Contact details typed by a customer who wants to be contacted are an intention; contact details copied from a profile are a record, and the two match against a closed-sales file at very different rates.

This is the part that gets missed when Meta leads are compared against leads from anywhere else in the budget, and it is worth being blunt about it: they are not the same kind of data and they should not be expected to behave the same way.

None of this makes the channel bad. A cheap lead with a sixty per cent chance of being reachable can still be a better buy than an expensive one that is always reachable — that is an arithmetic question, and it has a real answer. The problem is that nobody does the arithmetic, because the only numbers on the screen are lead volume and cost per lead, and both of those are flattered by exactly the mechanism that is weakening the leads.

There is a practical consequence for matching, too. Because the email on a Meta lead is the profile email, it is disproportionately likely to be a personal address the customer does not use for purchases — the one attached to a social account rather than the one attached to a bank card. A business that matches on email alone will see the channel underperform for reasons that have nothing to do with the advertising. Matching on the phone number, normalised to full international format, is usually the stronger key here, and matching on both, with either one sufficient, is stronger still.

If Meta leads match your sales file at a lower rate than every other channel, the first hypothesis should be the prefill, not the customers.

Meta deletes the lead before your sale closes

Lead data is available for download from Meta for 90 days after the form is submitted; after that it is gone, even though the campaign's lifetime results continue to count it.

This is the single most consequential fact about reconciling Meta lead ads, and it is documented in a help page most advertisers meet only after they have lost something. The download is not an archive. It is a window.

Take that last clause seriously — customer interest would have diminished — and then hold it against the sales cycles that actually exist. Ninety days is a long time for a gym membership and a short time for a conservatory, a hearing aid, a private school place, an implant consultation or a commercial roof. For any of those, the lead record expires on Meta's side while the deal is still open on yours. When the sale finally closes, the enquiry that produced it is no longer downloadable, and the revenue arrives with nothing to attach it to.

The fix is unglamorous and it is not optional:

• Connect a CRM integration or a webhook so that every lead lands in a system you control at the moment it is submitted, rather than waiting to be exported. Meta supports both, and this is what they are for.
• If leads are being downloaded by hand, download them weekly and keep the files. A quarterly export is already gambling with the oldest month in it.
• Store the Meta lead id alongside the contact details. It is the only value that unambiguously ties a row in your CRM back to a specific submission, and it costs nothing to keep.
• Keep the leads that were never contacted, and the ones marked junk. A lead file with the failures removed cannot tell you what the failures cost, and the failure rate is the number the whole exercise exists to measure.
• Record the form and the campaign each lead came from. Cost per sale by form is the comparison that eventually changes the account, and it cannot be reconstructed later from a list of phone numbers.

None of that is attribution yet. It is just refusing to lose the raw material — and a surprising share of the businesses that say Meta leads do not work for us are describing an export they stopped doing eight months ago.

Volume is a setting, and Meta will tell you so

Meta offers three instant form types, and the default is explicitly optimised for the number of leads rather than their quality — changing it is a one-field decision that most accounts have never revisited.

There is a form type selector in the instant form editor, and if nobody has touched it, it is set to More volume. That is not a criticism of the default; it is the sensible thing for Meta to ship. It is a criticism of leaving it there for two years and then judging the channel on the leads it produced.

An additional review screen that gives people a chance to confirm their information is, read plainly, an admission of everything above. The default form's speed is a feature bought at the price of deliberation, and the platform offers to sell some of that deliberation back to you. Whether the trade is worth taking is not a matter of opinion — it is measurable — but it is only measurable against closed sales. Switching to Higher intent will reduce lead volume and raise cost per lead. In a report built on cost per lead, that change looks like damage every single time, which is precisely why so few accounts keep it.

A change that improves revenue and worsens cost per lead will be reversed within a month by anybody reporting on cost per lead.

What to reconcile, and what to reconcile it against

Reconciling Meta lead ads requires the lead export — every lead, not only the ones the sales team liked — and the closed-sales file from whatever system records payment, joined on normalised phone numbers and email addresses.

From the Meta side, whether it arrives by CRM integration, webhook or download, each lead should carry: the submission timestamp; the campaign, ad set, ad and form it came from; the form type; every prefilled field the form requested, with the phone number and email intact and unmasked; the answers to any custom questions; and the Meta lead id.

From the sales side, export the closed and paid business for a window that extends well past the lead period — by at least the length of your longest normal sales cycle, and preferably by half as much again. Each row needs the customer's phone number and email as captured at the point of sale, the close date, and the value in the currency it was invoiced in. If the CRM carries a source field, bring it along, but treat it as a claim rather than a finding. It was typed by a person who was told something over the phone, and testing it is a large part of why the reconciliation is being done at all.

Two normalisation details decide most of the outcome here, and both are boring:

1. Phone numbers must be parsed to full international format on both sides, with the country inferred from the dataset rather than guessed per row. A leading zero dropped in one export and kept in the other is enough to send a whole channel's revenue to Direct / Unknown.
2. Emails must be lowercased and trimmed, and nothing beyond that should be normalised away. Stripping dots or plus-addressing is a provider-specific rule and applying it globally invents matches that are not there.

CloseRev takes the Meta lead export and the closed-sales file exactly as they come out of each system, normalises phone numbers and emails on both sides, and matches a paid sale to the lead that submitted the form. High-confidence matches count toward channel revenue; weaker candidates are shown to a person to accept or reject rather than being counted quietly; and sales that match no lead anywhere go to a Direct / Unknown bucket instead of being assigned to whichever channel looked most plausible.

What the reconciled account looks like

Once leads are matched to closed sales, the two form types and the several campaigns in a Meta account usually reorder themselves — and the campaign with the lowest cost per lead is frequently not the one paying for itself.

The figures below are invented, and deliberately round. The shape is the point, and the shape is one that turns up again and again in accounts running both form types side by side.

On the first three rows, the More volume form is winning by a distance and it is not close: three times the leads for slightly more money, at well under half the cost per lead. On the last four rows it is losing, and the mechanism is entirely visible — the cheaper leads are harder to reach, and the ones that are reached buy less often. The account that reports the top of that table monthly will keep spending into the left column and will have a perfectly good reason for doing so every time.

The unreached leads deserve their own line, because they split into two piles with different owners. Some were never going to be reachable: a stale number on a profile, a mistyped digit, someone who tapped through an ad in a feed and forgot. Those belong to the format, and the answer to them is the form type and the targeting. Others were reachable and were not reached — called once on a Friday afternoon, emailed from an address nobody monitors, or left in a queue for four days. Those belong to you, and the answer to them is a process change. The lead file cannot tell the two apart. The reconciled file, with the outcome next to the contact attempts, can.

Where the revenue goes when nobody joins the files

Unreconciled Meta lead revenue is not missing, it is misfiled — usually under branded search, direct, or a CRM source field somebody filled in from memory.

The path is dull and completely predictable. A customer sees the ad, submits the prefilled form, and gets a call the next day. They are half interested. A week later they look the company up by name, land on the website from a branded search, and book. Analytics records a booking from organic search. The CRM records the source as website. The Meta lead sits in the export, uncontacted status, never matched to anything, and ninety days later it is not even downloadable.

Or the shorter version: the salesperson closes the deal over the phone, opens the CRM, and picks a source from a dropdown they have never been trained on. Referral is the most-used option in most CRMs and the least reliable field in the entire dataset.

The result, repeated over a year, is a report in which Meta produces enormous lead volume and very little revenue, while branded search and direct look like the engine of the business. The recommendation writes itself, the lead ads budget gets cut, and the branded search volume quietly declines a quarter later for reasons nobody connects to the decision. This is the ordinary failure mode of a channel that creates demand and hands it to somebody else to close.

Cutting the channel that generates the enquiry and keeping the channel that harvests it is the most common self-inflicted wound in a marketing budget, and it always looks data-driven.

The honest Meta lead ads report

A defensible report for this channel shows spend, matched closed revenue, cost per closed sale by form and by campaign, the reach rate on the leads, and the leads that closed nothing, split into the format's failures and your own.

Five lines, and every one of them traceable to a file somebody can open:

1. Spend for the period, taken from Ads Manager, split by campaign and by form type.
2. Matched closed revenue: the invoiced value of every sale whose customer phone number or email matches a lead submitted in the period, with the lag between submission and close shown so the reader can see how much of the tail is still open.
3. Cost per closed sale and return on spend, by form type and campaign. This is the line that reallocates budget, and it is the one that cannot be produced without the join.
4. Reach rate — the share of leads that were successfully contacted at all — reported next to cost per lead, so that a cheap unreachable lead is never mistaken for a cheap lead.
5. Unmatched leads, listed and split: unreachable contact details on one side, contacted-and-lost on the other, and never-contacted as its own line, because that last one is a management problem wearing an attribution costume.

And the sales that matched no lead from any channel stay in Direct / Unknown, unassigned. Meta lead ads are a channel where the temptation to close that gap by inference is unusually strong, because the platform will happily model it for you and the modelled figure will be larger. A number you cannot trace to a specific form submission and a specific invoice is not evidence, however confident the interface that produced it looks.

The format is genuinely good. It removes the friction that kills mobile form fills, and for the right business at the right price it produces work that nothing else produces. But it removed the friction on both sides at once: it is now as easy to submit a form you did not mean to fill in as it is to submit one you did. The only instrument that can tell those two apart is the file of people who eventually paid you, and it lives in your systems, not in Meta's.

Questions people actually ask

Why do Meta lead ads produce so many leads that never answer the phone?

Because the form does not ask the customer to type anything. Meta prefills the contact fields from the person's Facebook or Instagram profile, so a lead can be submitted in two taps by someone who barely read the ad, using a phone number they last updated years ago. The submission is genuine; the intent and the contact details are both weaker than the volume suggests.

Is the phone number on a Meta lead ad verified?

No. It is whatever the person has on their Facebook or Instagram profile, and they can edit it before submitting. Meta verifies nothing about it for you. A prefilled number is convenient rather than confirmed, which is why match rates against a closed-sales file run lower for this channel than for a channel where the customer typed their number to get a call back.

How long do I have to download my Meta leads?

Ninety days from the moment the form is submitted. After that the lead is no longer available for download from Meta, even when your campaign's lifetime results still count it. Any business whose sales cycle runs longer than three months has to be capturing leads continuously — by CRM integration or webhook — because the record will be gone before the sale closes.

Should I switch from a More volume form to a Higher intent form?

Test it, and judge the result on closed revenue rather than on lead count, because lead count will certainly fall. Higher intent adds a review screen before submission, which Meta says is there to prevent submissions from people who are only marginally interested. That is a trade of volume for intent, and only a reconciliation against paid sales can tell you whether you took the good side of it.

What is the right cost metric for Meta lead ads?

Cost per closed sale, with the revenue that sale produced beside it. Cost per lead on an instant form measures the price of a tap on a prefilled form, which is the cheapest action in digital advertising and the one least connected to buying anything.

Found this useful?

• Share on X
• Share on LinkedIn
• Share on Reddit
• Share on Email