Why your ad platform reports 20x more conversions than your CRM received

We took a B2B lead pipeline apart recently and found that the ad platforms were reporting roughly 20x more conversions than the pipeline had actually received. Nothing was broken, and nobody had made a mistake. Every system was doing precisely what it had been told to do.

The problem sat in the CRM. Its lifecycle stages had stopped meaning anything, and it was feeding that back to the ad platforms as training data. When your CRM lies to your ad platforms, the platforms bid harder for the wrong people.

FIVE SIGNS, AT A GLANCE

	The sign	Why it misleads the platforms
1	Your ad platform reports far more conversions than you have real people in the CRM	A gap of multiples means one person is counted several times, and the bidding inflates to match
2	Almost everyone who fills in a form is marked qualified, often within a day	A label that applies to nearly everyone has stopped doing any filtering
3	No one records whether sales actually accepted the lead	Without a human yes or no, lead quality is never measured at all
4	Every lifecycle stage fires its own conversion	One person moving through the funnel looks like three separate wins
5	Your lead score is built entirely on browsing behaviour	It assumes buyers browse before they buy. In B2B, most do not

The tell that ties all five together: cheap leads and a quiet sales team, showing up at the same time.

Sign 1: Does your conversion count match your contact count?

Start with the crudest check available. Take the conversion count your ad platform reported for a given period, then count the real, distinct human beings who entered the CRM in that same period.

It is the gap of multiples that matters. That gap means the same person is being counted repeatedly, and the bidding is inflating to match a level of demand that does not exist.

Sign 2: What percentage of your form fills get marked qualified?

In the pipeline we audited, 8 in 10 form fills were marked qualified, most of them within a day of arriving.

Qualified had become a synonym for “submitted a form”. A label that applies to nearly everyone has stopped doing any filtering, and it teaches the platforms nothing at all about who is worth finding more of.

Sign 3: Is anyone recording whether sales accepted the lead?

Nobody was, and nothing looked wrong as a result.

Without a human yes or no recorded somewhere, you have no measure of lead quality. You cannot tell good traffic from bad, which means you cannot optimise, which leaves volume as the only thing to optimise towards.

Sign 4: How many conversions does one person generate?

Every lifecycle stage in this setup fired its own conversion event. One person moving normally through the funnel therefore looked like three separate wins, and the platform bought more of whatever caused them.

Stack that on top of a qualified stage that admits almost everyone and a quality measure that was never recorded, and a 20x gap stops being surprising. All three faults compound in the same direction.

Sign 5: Do your buyers actually browse before they buy?

This is the one most people miss, and it is where the audit produced its genuinely uncomfortable finding.

19 in 20 of the customers this business actually won had almost no website activity we could see before they bought. There was next to nothing there to score.

A lead score built on clicks and page views would have missed nearly every real customer they had. In B2B, who someone is beats what they browsed, and a behavioural score quietly inverts that.

How do you fix it?

This was subtraction, which is usually where Smart Simplicity actually shows up in practice. No new tool, no rebuild, no extra layer of tech. Four changes:

1. Qualified has to be earned. A lead only reaches the stage by clearing a score.

2. One human checkpoint. A single person reviews the promising leads and accepts or rejects them. That decision becomes the quality measure everything downstream learns from.

3. Automatic filtering at the front. Leads that clearly do not fit are removed before anyone spends time on them.

4. One clean signal out. The ad platforms receive a single conversion event instead of several overlapping ones.

Fewer “leads”, more actual conversations. That trade is almost always worth making, and the reporting only looks worse for as long as it takes people to stop comparing it to a number that was never real.

Final Takeaway: Cheap leads and a quiet sales team is the tell. When those two appear together, the problem is almost never what you are paying for clicks, it is what you are calling a lead. Fix the definition and the numbers get smaller, honest, and finally worth optimising against.