What the Score Is Actually Measuring
Meta grades every conversion event you send on a scale of 0 to 10 and calls it Event Match Quality. The score is set by how many of its fifteen customer-information parameters your event carries, and whether they match a real Meta user.
Five of those parameters arrive on their own. The client IP address and user agent come with the request. The two Meta browser identifiers, fbc and fbp, come from the ad click. Email is the fifth, and only when your own landing page collected it. Any competent server-side setup carries these five without you thinking about it.
The other ten are phone, first and last name, date of birth, gender, city, state, zip, country, and an external ID. Every one of them requires somebody to ask a customer a question and write the answer down. On an ecommerce store, that somebody is the store, and the answers are already in the database when the purchase event fires. On an affiliate offer, that somebody is the vendor.
Every green checkmark in the setup is telling the truth; none of them are measuring the thing that actually sets the score.
The Four Seconds Where It Leaks
Follow one buyer through a normal network offer and you can watch the identity leave.
They see the ad and click. Your tracker gets both Meta browser identifiers, the IP, and the user agent. Four parameters, and you have not done anything yet.
They land on your page. This is the only moment in the entire journey where you can ask them a question. If your page collects an email, you now have five. If it collects a phone as well, six. If it collects nothing, you are still at four.
They click through to the offer. The domain changes. Everything after this point happens on the vendor's property.
They fill in the order form: first name, last name, billing address, city, state, zip, country, card details. Ten fields, give or take, entered into somebody else's database. Time yourself filling one in and you will land somewhere around four seconds of typing, which is the entire width of the moment your Event Match Quality score is decided in.
The sale completes and the network fires a postback to your tracker. It carries the click identifier and a payout amount. It does not carry the name. It does not carry the address. It was never going to.
Ten of the fifteen match keys are entered into the vendor's form in about four seconds, and the postback that comes back to you carries none of them.
The Math Nobody Puts on a Slide
Here is what the gap costs in the only terms that matter.
A conversion Meta cannot confidently match to a user still counts in your reporting. It just teaches the model nothing, because the model never established who bought. You paid for the click, you paid for the traffic, you generated the sale, and then you handed the platform an event it could only half read.
Two campaigns can spend the same money, produce the same sales, and send events of very different quality. The one sending fifteen parameters gets a training set the algorithm can actually use. The one sending five gets counted. Over a few weeks of learning that difference stops being a scorecard number and starts being your cost per acquisition, which is why I treat Event Match Quality as a performance metric rather than a health check.
And it compounds in the wrong direction as browser identifiers keep getting shorter-lived. The five free parameters are the ones most exposed to that decay. The ten you have to earn are the ones that hold.
Getting the Fields Back Before the Handoff
You cannot reach into the vendor's checkout, and you should not try. What you can do is move the identity collection earlier, onto the one page in the journey that belongs to you.
That is what a pre-sell page, a quiz funnel, a lead magnet, or a simple email gate is actually doing for your tracking, whatever else you built it for. It converts an anonymous click into a known person before the handoff happens. Once you hold an email against a click identifier, you have a spine, and anything else you asked for, first name, country, zip, sits on it.
Then the second half has to work, and this is where most setups quietly fail. The identity has to survive until the network postback lands hours or days later, get joined back to the right click, and be hashed and attached to the conversion event. If your tracker stores the opt-in and then forgets it by the time the sale reports, you did the hard part and threw it away at the last step.
Three questions separate a 5-parameter event from a 15-parameter one, and the middle question is the one most setups fail without noticing.
Why This Got Worse, Not Better
Two things changed underneath affiliates over the last few years and they pulled in the same direction.
Browser-side identifiers kept getting shorter-lived and easier to lose, so the share of conversions matchable on a click identifier alone kept falling. That was the parameter most affiliate setups leaned on hardest, precisely because it was free.
At the same time the platforms leaned harder into their own modeling. When a platform cannot match your event to a user, it does not throw the conversion away, it estimates around it. That estimate is built from advertisers who did send identity. So the buyer with the thin event is not just learning slower, their delivery is being steered by a model calibrated on somebody else's data.
Neither of those is a scandal. Both mean the free five are worth less every year and the earned ten are worth more.
So What Do You Do About It
Pull up your Event Match Quality score before you change anything, and count how many of the fifteen your events actually carry. If it is five or fewer, stop auditing your Conversions API integration, because the pipe is almost certainly fine. Go look at the last page you control before the vendor takes over, and ask what you are collecting there. That page is the entire width of your opportunity.
Then make sure what you collect survives to the postback. ClickerVolt was built around exactly that join: identity captured on your page is stored against the click and forwarded with all fifteen Meta parameters when the conversion reports, days later if that is how long it takes. See how the full-parameter signal works. Whatever tracker you use, the takeaway holds: you are never going to own the checkout, so own the page before it.
