Skip to main content
FHAVI

Free resource

One order, two senders, side by side

Server-side tracking is only an improvement if the browser and the server agree on what happened. Otherwise you have two systems describing one sale, and the platform has to guess. This is how you check in fifteen minutes, with a single order you cancel afterwards.

Before the test

Set up to see both halves at once

You need three things open: Events Manager with the Test Events tab and a test code copied into your tracking settings, the storefront in a private window with extensions off, and somewhere to write the comparison down.

The private window matters. An ad blocker will remove the browser half and leave the server half, which looks exactly like a working server-side setup and a broken pixel — a false conclusion people draw all the time.

Place one order, on your own phone number, cash on delivery, then cancel it in admin when the test is done. Do not use a live customer order and do not fire a fake purchase event into a production pixel to “test” it.

The sheet

What to record for the same order

01

The identifiers

  1. Order number — the id that should appear in both events, unchanged.
  2. Browser event id and server event id, read from Events Manager or from your request logs. If they differ, stop: everything else will be wrong.
  3. fbp and fbc cookies present on the order, and whether the server payload carried them.
  4. Hashed phone and email present in the server event — check the payload, not the intent.
  5. Currency and value in both events, and whether they match the order total to the paisa.
02

The result in Events Manager

  1. One purchase row, or two. One with two sources is the pass condition.
  2. Source labels shown: Browser, Server, or both.
  3. Deduplication status, if the column is visible for that event.
  4. Event match quality score, if the dataset shows one.
  5. Time received — a server event arriving hours after the browser event usually means a queued or retried send.
03

The failure patterns

  1. Two events, two ids → duplication. Revenue overstates, and campaigns that do not convert look profitable.
  2. One event, server only → the browser pixel is blocked or missing. Server-side is doing its job, but match quality is lower than it should be.
  3. One event, browser only → the server send is failing silently. It will look fine for months and stop when traffic drops.
  4. One event, both sources, values differ → a payload bug, usually shipping or a discount applied on one side only.
  5. Nothing at all → the test code is not active, or you are looking at the wrong dataset.

After the test

What to fix, in order

04

Priority

  1. Fix duplication first. Overstated revenue corrupts decisions in a way that a gap does not, because it funds bad campaigns confidently.
  2. Fix a silent server failure second — the one where the code runs, returns success, and Meta rejects the event.
  3. Improve match quality third: add the missing identifiers rather than restructuring what already works.
  4. Re-run this same test after any checkout change, theme change, domain change or platform migration.
05

Keep the evidence

  1. Screenshot the merged event with the order number visible, and keep it. It is the fastest way to prove your numbers are trustworthy to a client, an agency or a new hire.
  2. Record the date and the value, so a later discrepancy can be compared against a known-good state.

More resources

Keep going

The rest of the library

Every checklist here is free and ungated. The research pages carry the reasoning behind them.

All resources · Research library

Both halves sending and still not matching?

A single test order is enough for us to tell you which half is wrong. Send the order number and the platform and we will be specific.

Prefer email? Write to md@fhavi.com