Conversion Tracking Audit: The Pre-Q4 Checklist for Reliable Data
Before Black Friday and the holidays, make sure every conversion is counted once, with the right value and valid consent. The full checklist and a practical test protocol.

A conversion tracking audit checks that every conversion (purchase, form, call) is recorded exactly once, with the right value and valid consent. Before the year-end spending peak, it is the most profitable check you can run: Google Ads and Meta algorithms amplify whatever data you feed them, right or wrong.
Here is the checklist we use before every high season, plus a test protocol that will not eat your whole week.
Key takeaways
- Audit in late October, not the day before Black Friday: fixes need time, and so do the algorithms that learn from them.
- Check five things in Google Ads: primary actions, duplicates, values, counting and attribution windows.
- Verify Consent Mode v2 and your cookie banner.
- Reconcile Google Ads, GA4 and your back office: a gap is normal, an unexplained gap is not.
- Freeze tracking changes for the two weeks before the peak.
Why bad tracking gets expensive in Q4
Automated bidding (Smart Bidding in Google Ads, delivery optimisation in Meta) decides how much to bid in every auction based on the conversions it sees. Count a conversion twice and the algorithm thinks that traffic is twice as profitable, so it bids higher. Send a wrong value and a target ROAS strategy pushes the wrong products. Lose a conversion and it cuts campaigns that were actually working.
For most of the year, these errors cost money. In Q4, when budgets and CPCs climb, they cost a lot more, and there is very little time left to fix them.
The Google Ads conversion checklist
| Checkpoint | What good looks like | Red flag |
|---|---|---|
| Primary actions | Only your real business goals | Add-to-cart or page views set as primary |
| Duplicates | One measurement path per conversion | An unusually high conversion rate |
| Values and currency | Real order value, correct currency, one tax rule everywhere | The same fixed value on every order |
| Counting | “Every” for sales, “One” for leads | More reported leads than real prospects |
| Attribution | Known, documented windows and model | Settings changed recently for no clear reason |
Primary and secondary actions
Only primary conversion actions drive bidding. Make sure they are your real business outcomes (purchase, quote request, qualified call) and that micro-conversions (add-to-cart, page visits, button clicks) are set to secondary. One micro-conversion left as primary skews the entire optimisation.
Duplicates
Look for actions that record the same event twice: a Google Ads tag and a GA4 import both active, a tag installed both in the site code and in Google Tag Manager, a thank-you page that gets reloaded or revisited. The rule is simple: one measurement path per conversion.
Values, counting and attribution
In e-commerce, the value sent should be the real order amount, in the right currency, with one consistent rule (with or without tax, with or without shipping). For lead generation, assign different values when some enquiries are worth far more than others. Set counting to “Every” for purchases (each order counts) and “One” for leads (a prospect who sends three forms is still one lead). Finally, write down your conversion windows (30 days after a click by default in Google Ads) and the attribution setting of each Meta ad set.
Consent Mode v2 and your cookie banner
For visitors in the European Economic Area, Google requires Consent Mode v2 to keep using their data for measurement and personalised advertising: the ad_user_data and ad_personalization parameters sit alongside ad_storage and analytics_storage. Check that:
- The banner appears before any advertising cookie is set, with a “Reject” option as easy to reach as “Accept”.
- Consent signals actually reach your Google tags (you can see them in Tag Assistant).
- Your chosen mode matches your consent platform: basic (no tags before consent) or advanced (cookieless pings that allow conversion modelling).
- A refusal is truly respected: no advertising tag sets a cookie after someone clicks “Reject”.
If consent is not passed correctly, part of your conversions is neither observed nor modelled, and your audience lists shrink exactly when you need them most.
Reconciling Google Ads, GA4 and your back office
The numbers will never match perfectly, and that is expected. The usual reasons:
- Attribution: Google Ads claims conversions that follow its clicks, while GA4 spreads credit across every channel.
- Dates: Google Ads reports a conversion on the day of the click; GA4 and your CRM report it on the day it happened.
- Consent and browsers: cookie refusals, ad blockers and browser privacy restrictions.
- Cancellations and returns: your back office reflects them, ad platforms rarely do.
The method: over the same 30-day period, compare the number of orders or leads in your back office, in GA4 and in Google Ads. As an indicative benchmark, a gap of 10% to 20% between tools is common. A gap that suddenly shifts from one month to the next, or platforms reporting more sales than your till, points to a problem to fix before the peak.
The test protocol
- Google Tag Manager Preview mode: walk through the full funnel (product page, cart, checkout, confirmation; or form and thank-you page) and check that each tag fires once, at the right moment, with the right variables.
- Tag Assistant: confirm that the Google tag receives the consent state and that the conversion event carries the value, currency and order ID.
- Meta Events Manager: use the Test Events tool to check the Pixel and the Conversions API, and above all deduplication (the same event_id on the browser and server sides).
- A real conversion: place a genuine order or submit a genuine form, then find it in each tool the next day. Remember to cancel or exclude it afterwards.
- Diagnostics: review conversion action status in Google Ads and Event Match Quality in Meta.
The freeze list: what not to touch before the peak
Once tracking is validated, set a freeze date, ideally two weeks before Black Friday (27 November 2026). Until the peak is over, avoid:
- creating, deleting or promoting a conversion action to primary;
- changing attribution windows or models;
- switching consent platform, theme or checkout, let alone migrating the site;
- publishing a Google Tag Manager container version that has not been tested;
- abruptly changing CPA or ROAS targets to “make up” for an unexpected number before checking tracking first.
If something breaks anyway (a broken tag, site downtime), Google Ads lets you apply a data exclusion so Smart Bidding ignores the affected period. Log every incident with its date: that record will also help you analyse the season.
Get your tracking checked before the peak
Tracking is the foundation of our Google Ads and Meta Ads management: we verify it before we touch a single bid. If you would like a second pair of eyes on your conversions before Black Friday, book a free audit with one of our specialists. The call is free, with no obligation, and you leave with a prioritised list of fixes.
Frequently asked questions
Q.01When should I audit conversion tracking before Black Friday?
Ideally in late October or early November. You need time to fix, test and let bid strategies settle before the peak, with a freeze on changes two weeks beforehand.
Q.02Why don't Google Ads and GA4 show the same conversions?
They use different attribution and dates: Google Ads reports conversions on the click date and claims the credit, while GA4 spreads credit across channels and records the conversion on the day it happens.
Q.03How do I check that Consent Mode v2 is working?
Open Tag Assistant and check that the Google tag receives the consent state, then confirm that once a visitor declines, no advertising tag sets a cookie. For why it is required and how to install it, read our Consent Mode v2 guide.
Q.04How can I spot a duplicated conversion?
An unusually high conversion rate is the first sign. Then check that the same conversion is not recorded by both the Google Ads tag and a GA4 import, or by two installs of the same tag.


