Marketing Measurement

Meta received 640. You sold 1,000. Why?

Your order system shows 1,000 website purchases. Meta Events Manager received 640 events. Here is how to read that gap.

Hardal
3 min read
Meta received 640. You sold 1,000. Why?

Your order system shows 1,000 website purchases this month. Meta Events Manager received 640 purchase events.

Both of those numbers can be true at the same time. Events Manager is not a second copy of your order book. It is a count of the purchase events Meta actually received.

This is easier to see with a kitchen in mind. Waiters take 100 orders, but only 64 tickets reach the kitchen. The kitchen never heard about the rest, so it cannot cook them, learn from them, or plan around them.

Why the signal goes missing

Browser-only tracking fails in the same quiet way. Ad blockers stop the pixel from firing. Browser privacy rules shorten cookie lifetimes and strip click IDs, so some events arrive without the details Meta needs to match them. Slow pages mean the customer has left before the event is sent. The sale still happens in your own system. The signal never arrives.

Consent is a different matter. When someone declines, you must not send their data. Server-side tracking is not a way to recover those events, and it should not be used that way under rules such as GDPR and KVKK. Consent choices reduce what you can send, and those choices you have to respect.

The events you do send are the only ones Meta can learn from. The platform bids on the customers it can see. If it received 640 purchase events, it is training and bidding on 640, not on 1,000.

What server-side tracking actually does

Server-side tracking usually sits alongside the browser pixel, not in place of it. That is useful, because it can recover events lost to blockers, browser limits and slow pages. It also creates a new risk. If the pixel and the server both send the same purchase, Meta may count it twice unless the two events share an event ID. The gap you were trying to close then turns into double counting.

Event Match Quality

Once events arrive, the next question is whether Meta can match them to a person. Event Match Quality is Meta's score for this, from 0 to 10. The score depends on the customer details you send with each event, such as a hashed email, a hashed phone number, IP address, user agent and the click ID. Stronger match quality helps the platform attribute the event and find more people like that customer. Weaker match quality leaves the event underused, even if it arrived.

Growth teams can start with three questions. How many of our real website conversions does each platform receive? Are we sending them from our servers as well as the browser, with a shared event ID? Can the platform match those events to people?

A conversion you count but never send is invisible to the algorithm spending your money.

Today's small task

Compare last week's purchase events in Meta Events Manager with the website orders in your own system. Use the same time zone, leave out cancelled orders, and write the gap as a percentage. Leave app and phone orders out of this comparison, so you are not mixing two different pictures.

If you want help closing that gap with first-party, server-side data, start at hardal.ai.

Hardal is the first-party data and agentic analytics platform. We help enterprises send accurate signals to their ad platforms and grow revenue, using server-side data from their websites, mobile apps and AI agents.