A broken tag is worst on your busiest week.
Nine checks protect a peak period. Submit a real test purchase and confirm the conversion arrives with the right value. Check nothing is counted twice. Confirm the consent banner is not blocking tags for a share of visitors. Compare platform conversions against orders in your own system. Check mobile separately from desktop. Confirm the thank you page cannot be reached directly. Check that refunds are handled. Confirm the enhanced or offline conversion feed is still flowing. And take a screenshot of a normal week now, so you have something to compare against later.
Tracking fails after ordinary changes
Nobody breaks tracking on purpose. It breaks when a theme is updated, a plugin changes, a checkout is redesigned, a consent tool is installed, or a developer tidies up something that looked unused.
All of those are routine, which is exactly why this needs checking on a schedule rather than when something looks wrong. By the time a number looks wrong, the period it describes is over.
The test that matters most takes ten minutes
Buy something on your own site, or submit a real enquiry, and follow it all the way through. Does the conversion appear in the advertising platform, does it carry the right value, and does it appear exactly once.
Most tracking problems are found by this single test, and almost nobody does it. It is more reliable than any preview mode, because it exercises the same path a customer takes rather than a simulated one.
The rest of the list
None of these take long, and each one has ruined somebody's quarter.
- Double counting. A tag installed twice, or fired by both a plugin and the tag manager, inflates everything and makes the advertising look better than it is.
- Consent. If a banner blocks tags until acceptance, a share of conversions are never recorded, and that share changes whenever the banner does.
- Reconciliation. Compare the platform's conversion count against real orders in your own system for the same week. They will not match exactly, and the size of the gap is the thing to know before it matters.
- Mobile separately. Some failures only occur on one device type, and mobile is where most of the traffic is.
- The thank you page. If it can be reached or refreshed directly, it will be, and every reload is another conversion that never happened.
- Refunds and cancellations, so the value you optimise towards is the value you kept.
- Any offline or enhanced conversion feed, which quietly stops when a credential expires.
Record what normal looks like
Before the busy period, save the numbers from a typical week: conversion rate by device, the gap between platform conversions and real orders, and the usual value per order.
That snapshot is what makes a later problem diagnosable. Without it, an odd looking December has nothing to be odd compared to, and every explanation is a guess.
Why this matters more at peak than at any other time
Automated bidding learns from conversions. During the weeks when you spend most, wrong conversion data teaches the system the wrong lesson at the highest possible cost, and it keeps applying that lesson afterwards.
So the damage is not limited to the reporting. It is in the decisions the account makes for weeks after, which is why tracking problems are the most expensive kind of quiet failure in paid media.
When did you last test your own checkout
Not looked at a dashboard. Actually bought something and followed the conversion through. If it has been more than a quarter, this afternoon is a good use of an hour.
If you want a second pair of eyes on the setup before your busiest weeks, send us access through the enquiry form. We will run the list and tell you what we find, including when the answer is that everything is fine.
Questions we get asked.
How do I know if my conversion tracking is working?
- Submit a real purchase or enquiry yourself and follow it through. Confirm the conversion appears in the platform, carries the right value, and appears exactly once. That single test finds most problems and almost nobody runs it.
Why does tracking break?
- Ordinary changes. Theme updates, plugin changes, checkout redesigns, consent tools and developer cleanups all break tags without anybody intending to. That is why it belongs on a schedule rather than being checked when a number looks odd.
How close should platform conversions be to real orders?
- They will never match exactly, because of attribution windows, blocked tags and cross device journeys. What matters is knowing the size of your usual gap, so a change in it becomes visible.
Does a consent banner reduce my conversions?
- It reduces the conversions you record, which then affects what automated bidding learns from. The number to watch is the share of visitors who never accept, and whether that share changed when the banner did.
Why is broken tracking worse during a busy period?
- Because bidding learns from conversions. Wrong data during your highest spending weeks teaches the system the wrong lesson at the highest cost, and it keeps acting on that lesson for weeks afterwards.
