Positive Control. Any conversion tracking setup either holds this one or it does not, whoever built it.
Positive Control
You checked the numbers at 2 AM. You checked them again with the first coffee, hunting for whatever went wrong overnight. That is dread rather than curiosity, and it means part of you already knows the dashboard can lie to you for weeks.
The propertyPositive Control is every conversion watched step by step, browser to our layer to what we sent to what the network confirmed to the CRM, monitored and alerted automatically.
Why looking at it is not the same as watching it
In Ads Manager, a bad day on the ads and a dead tracking pipeline look exactly the same. Conversions are down. Cost per acquisition is up. There is nothing on the screen that distinguishes an audience going cold from an event that quietly stopped firing after a page change.
You cannot tell the difference by looking, and that is why looking fails as a control. You can stare at that screen every morning for a month and still be reading a story about your creative when what you are actually looking at is a broken tag.
You may already have a version of this in mind, and it is a reasonable one. Somebody fires a test lead and watches it land in the platform, and on that evidence the setup gets signed off as healthy. The check is borrowed from the lab, where you prove an instrument can register a result before you trust it to report one, and it is worth running. It answers a narrow question, though. The path was open at the moment you looked. The failures that cost you start after you look, and they keep quiet, so a check that happened once leaves every day after it unattended.
Why I built this
On 6 May the web lead event on an account stopped firing. It was 12 days before anybody noticed and 28 before it was fixed.
We were the tracking vendor on that account. There was no monitoring in place, and to be straight about it, none had been proposed either. Nobody refused it and nobody was careless. It simply was not anybody’s job, which is the state almost every account is in.
The person who caught it was the client’s own media buyer, working through the numbers by hand. For May, he found one campaign showing 79 conversions in the platform against 254 in the CRM, and across the whole account a cost per lead reading about $657 when the real figure, counting primary leads only, was about $244. Counted gross, across every lead that arrived, it was nearer $182. The $244 is the one to hold against your own cost per prime lead.
Those numbers are how the failure came to light, 12 days late, through a person doing by hand the reconciliation that should have been running on its own. The proof on this page is the 12 days themselves. That is the Parity Lock comparison arriving far too slowly to be any use.
For those 12 days he had been carrying something worse than a wrong number. He had been making decisions on it the whole time, with no way of knowing whether the campaigns he had paused were failing or were converting perfectly well into a CRM the platform could no longer see. Every optimisation he had made in that window was suspect, and he could not tell which ones.
The same account had done this before and we had missed that one too. A campaign went blind on 18 February when a goal change excluded the form submit action, and it stayed blind for about seven weeks, until a media buyer found it by hand in April. Two incidents, one account, and we were the tracking vendor across both of them. That is the argument, and it costs me something to make it: watching is a job, and if it is not somebody’s job then it is not happening, ours included.
Neither failure announced itself. The webhooks kept arriving exactly as they were supposed to and every dashboard looked normal, which is what a silent failure looks like from the inside. There is no way to catch a silent server side failure without something actively watching for it, and until something is, the clock starts when the tag stops and it does not stop until a human happens to look.
What it looks like when it is on
The guarantee is that a conversion is followed the whole way, hop by hop, and that a hop failing raises an alarm on its own rather than waiting for somebody to go looking. A monitor would have flagged either of those two incidents on day one. Both instead took a person, by hand, and one of them took seven weeks.
Positive Control proves the pipeline is intact. Parity Lock proves it is right. You can have perfect control over the wrong set of events.
The Complete Count is the right events and essentially all of them, by design. Positive Control is each one proven to have arrived, with an alert the moment one does not.
Would we notice if our tracking stopped working?
Usually put to me as“we would notice.”
A total stop you would notice, yes. Conversions at zero on a Monday morning is the kind of thing somebody raises in the first hour. The expensive failure is the partial one: a third of your conversions stop arriving and the rest keep coming, so the account still reports leads and the graph still moves. The whole thing reads as a soft week. Soft weeks happen for a dozen ordinary reasons, so nobody escalates. Meanwhile the ad platforms are treating the missing third as people who did not convert and adjusting who they show your ads to accordingly, which makes the following week genuinely worse and confirms the story that it was a rough patch. Noticing a 30% drop takes something comparing what happened against what arrived, every day, on purpose.
One question sizes this up on your own account: if your conversions stopped arriving this afternoon, what would tell you, and how long would it take? If you would rather we looked first, send us your busiest landing page on each traffic source you run and we will write up what your platforms cannot see.
Find what your platforms can't see