Meta Says 40 Leads, Your Register Shows 22 — Which One Do You Run the Business On?

Forty on the dashboard, twenty-two in your register, and someone in the room insisting the other number is wrong. Here is the calm version: neither is lying. Meta reports the conversions it attributes and models inside its own reporting window; your records count the enquiries that actually reached you. The two are measuring different things, so they will never tie out to the rupee — and the gap got wider once Meta tightened how it counts. The short version: run the business on your own number — the register, the spreadsheet, the CRM, wherever real enquiries land — and use Meta's number only to steer Meta, and only when Meta can actually see your conversions.

Why don't Meta and my own records ever match?

Four reasons, and not one of them means your ads are broken:

• Attribution windows. Meta only credits a conversion that happens inside its window after an ad interaction. A study-abroad consultancy in Kottayam might get an enquiry today from someone who tapped the ad three weeks ago — real in the register, invisible to Meta, because the click and the form fell on opposite sides of the window.
• Modelled conversions. After privacy changes cut off a slice of tracking, Meta estimates some conversions statistically rather than observing them. They show up in Ads Manager, but they were never individual, verifiable events you can point to by name.
• Deduplication and cross-device. A villa buyer taps your ad on a phone and later calls from a landline; Meta may merge or split those touches differently than you do when you tidy up your enquiry list.
• Different definitions of "a lead." Meta counts a form submit. Your record might only count an enquiry after a duplicate is removed, a spam entry is stripped out, or someone has actually confirmed the person is real.

Put together, these four are enough to turn 40 into 22 without anyone doing anything wrong.

Which number do I run the business on?

Use each for the job it's good at:

• Your own record is the source of truth for the business — real people, real money. Make budget and profitability calls on this, whether it lives in a CRM, a spreadsheet, or a WhatsApp thread you scroll back through.
• Meta's number is a steering signal for the algorithm — it's how Meta decides who to show your ads to next. It's directionally useful inside the platform even when it doesn't reconcile with what reached you.

The mistake runs both ways: trusting Ads Manager as gospel for revenue, or throwing it out entirely because it doesn't match. It's a claim, not a receipt — but it's the claim Meta optimises on, so you can't just ignore it either.

Did my results actually drop, or did Meta just change how it counts?

Often it's the counting — and this one catches a lot of owners out. Meta reclassified what counts as a click: click-through now counts link clicks only, and non-link actions like likes, shares and saves moved to a separate "engage-through" bucket (reconfirm against Meta's current help docs). It also leans more on modelled data than it used to. The result is that reported numbers can fall even where real performance didn't change at all. So if your Ads Manager numbers seem to have fallen off a cliff on a particular recent date, with nothing changed at your end, suspect a counting change before you suspect your ads. The enquiries in your register are the reality check: if they held steady while Meta's number dropped, the ads are fine — the scoreboard was rewired.

The gap has a direction — and the direction is the diagnosis

There's always some gap; that's expected. What isn't normal is a large, one-directional gap, and which way it leans tells you what's actually wrong.

When Meta reports far more than reaches you (the classic 40-versus-22), it's usually one of three things:

• Meta is counting low-quality or duplicate form fills that your records discard — a goal problem, where the account is optimising for volume rather than real enquiries.
• The data connection between your enquiries and Meta is broken or incomplete, so Meta is guessing more than it's measuring — a tracking problem.
• A long sales cycle sits outside Meta's window — a reporting limit, not a lost lead.

When your records show more than Meta does, the leads are real but Meta couldn't see them — they came in through a channel it isn't tracking, or the conversion happened well after the window closed. That's a measurement blind spot, not a performance win to celebrate. Either way, read the direction first; it points you at the fix.

How do I make the two line up better?

You'll never get them identical, and chasing that is a waste of an afternoon. But you can close the gap and — more importantly — feed Meta a better signal:

1. Send your real conversions back to Meta through the Conversions API or offline/CRM events, so it optimises toward the enquiries that actually became customers, not just the ones that filled a form.
2. Keep your attribution settings consistent month to month, so you're not quietly comparing two different windows and calling the difference a trend.
3. Agree on one definition of "a lead" across Meta and your records, so both are at least trying to count the same thing.
4. Tie Meta spend to what closed — cost per real customer, pulled from your own record. That's the only view that answers the question that actually matters, and it's the one most accounts never build.

None of this runs itself. Someone has to decide which outcome is worth chasing, keep the connection alive, and check it hasn't silently broken — and it comes with no guarantee that the numbers converge. You own that direction; the tools only execute it. But that last step — spend traced to closed revenue — is the whole game, and it's exactly what most owners never get to.

So what is "attribution," really?

Attribution is Meta's best guess at which ad caused which result, inside a set time window. "Modelled" conversions are the ones it estimates rather than watches happen. Both are genuinely useful for steering the algorithm, and both are misleading the moment you treat them as your books. Keep the two jobs separate — Meta's number to point the platform, your own record to run the business — and the 40-versus-22 argument stops being a fight and becomes a diagnosis.