NOTE-024 · FIELD NOTE FILED 11 NOV 2026 · REVISED 11 NOV 2026 F-101 F-102 F-302 F-201 F-205
Tracking checks to run before Black Friday: the two-week list.
ABSTRACT — Black Friday is 27 November 2026. Two weeks is enough to test one order per payment method, fix faults in the order they distort bidding (missing or double purchases, wrong value, consent, lost source), freeze tag and app changes on 23 November, and name who watches each platform over the weekend. Rebuilds wait until December.
Baris Asa, tracking specialist, London · Filed 11 November 2026 · Last checked: November 2026
Black Friday is Friday 27 November 2026. Test one real order per payment method by 20 November. Fix faults in the order they distort bidding: missing or double purchases, wrong value, consent, then lost source. Freeze tags, apps and conversion settings on 23 November. Name who watches each platform from 27 to 30 November. Rebuilds wait until December.This note is about timing and order: what fits in the fortnight, what to fix first, and what to do if something breaks mid-sale. What each check involves in full is the ecommerce tracking audit checklist. What an audit report should hand back is a separate note.
Can I check anything today, before a test order?
Yes. Compare last month’s GA4 purchase revenue with the store’s own sales report for the same dates. Some gap is normal (refunds, consent, ad blockers). GA4 well ahead of the store usually means purchases counted twice, on file as F-101 : start there. The reasons it runs ahead are in GA4 revenue higher than Shopify.
On Shopify, also check where the purchase tracking lives now. Shopify switched off Additional scripts on the Thank you and Order status pages for every plan by 26 August 2026, so code left there no longer runs. What replaces it: Shopify checkout.liquid and Additional Scripts are gone.
What fits in two weeks, and what doesn’t?
Two weeks fits testing and small fixes. It doesn’t fit rebuilds.
Fits: test orders; removing a second purchase sender (an app and a Google Tag Manager tag both sending the sale); fixing a missing or reused order ID, a wrong value, or a payment provider that strips the sale’s source; checking the consent default and the banner.
Doesn’t fit: a move to server-side tracking, a new consent platform, a change to which conversion Google Ads or Meta optimises for (unless the current one is wrong), a theme rebuild, or a new payment provider. Each changes more than it tests, and none can be proved on enough real orders before the sale.
If you’re reading this after 18 November: do the test orders and the freeze, and leave everything else for December.
Week one (16 to 20 November): test, compare, list
Test on the Monday or Tuesday, so a fix can be re-tested before the week ends.
- Place the test orders. One per payment method (card, PayPal, Klarna, whatever the checkout offers), plus one with the banner accepted and one rejected. Use a low-value product, note every order number, and refund them the same day. A refund doesn’t remove the purchase from GA4 or Meta, so the order numbers are what let you explain them in the reports later.
- Open Tag Assistant or GTM Preview before you buy. That is what makes the order show in GA4 DebugView, and it is where you see the Google Ads conversion tag fire.
- Find each order. GA4: DebugView now, the Transactions report the next day, one row with the store’s order number and value. Meta: Events Manager, one Purchase with the same value. Google Ads: a test order without an ad click won’t appear in the conversions columns, so the check is the tag itself, firing once with the order number, value and currency.
- Read the order number. If DebugView shows two purchase events for one order, or the Transactions report lists it under a blank ID or one that isn’t your order number, the purchase is counted wrong. GA4 removes a repeat with the same ID; Meta needs its own matching
event_id. How each platform removes duplicates, and why a blank ID breaks it: GA4 purchase event fires twice and Meta CAPI deduplication check. - Check the return from each payment provider. The order should keep the source of the link you arrived on, not show as a referral from the provider.
- List what changed since anyone last tested. New apps, a theme release, a new GTM tag, a changed banner. Recent changes are where new faults come from.
| Order and method | GA4 | Google Ads tag | Meta |
|---|---|---|---|
| #1041, card | 1 purchase, value matches | fired once, ID matches | 1 Purchase |
| #1042, PayPal | 1 purchase, source lost | fired once, ID matches | 1 Purchase |
| #1043, banner rejected | no cookied purchase | consent state correct | per banner setup |
A tick without an order number behind it is not evidence. If a row shows a mismatch nobody can explain by Wednesday 18 November, hand it to whoever owns the tags, or to an auditor, while there’s still time to fix and re-test.
Which faults do I fix first?
Fix in the order they distort decisions. Google Ads and Meta bid on the purchases they’re sent, so faults in that signal come first.
- Purchases missing or counted twice. Missing purchases starve bidding; doubled ones inflate it. F-101 , and between Meta’s pixel and server events F-102 .
- The wrong value. Tax, shipping or the wrong currency in the purchase value skews value-based bidding. F-302 .
- Consent wired wrong. A default set after the tags fire, or an update that never arrives. F-201 . Set right, accepted visitors are measured in full and, in advanced consent mode where traffic clears Google’s thresholds, Google can model part of the rest. How to check: how to check Consent Mode v2 is working.
- The source lost at the payment provider. The sale still records, but GA4 credits it to PayPal or Klarna instead of the campaign. F-205 . The fix: payment gateway showing as referral.
Anything below that line (naming, unused tags, report tidying) goes on the December list.
Week two (23 to 26 November): freeze and re-test
Monday 23 November: freeze. From here, no GTM publishes, no new apps, no theme releases and no conversion setting changes, unless the change fixes a fault from week one. Tell the agency, the developer and whoever installs apps. A freeze nobody knows about isn’t one.
Before the freeze, write down the live GTM version number. Google Tag Manager keeps every published version. To go back, open Versions, pick the last good one, choose Publish from its actions menu and confirm. It takes a minute if you know which number to pick.
Thursday 26 November: one last test order per payment method. Same table. If every row matches, the setup going into the sale is the one you tested.
Should I change Google Ads settings before Black Friday?
That decision belongs to whoever runs the ads. The tracking side is narrow: don’t change what is being counted in the week before.
Google says a bid strategy goes into Learning when the conversion goals it optimises change significantly, and that Learning never lasts more than 7 days. Budget and target changes don’t trigger it. So a conversion action swapped in on 24 November may still be in Learning on the day of the sale. Seasonality adjustments are a bidding tool for short events of 1 to 7 days; whether to use one is your PPC lead’s call.
If you’ve just fixed a counting fault and the numbers look worse, that is usually the fix showing the real figure: Google Ads performance worse after fixing conversion tracking.
Who watches what over the weekend?
Name one person per platform for Friday 27 to Monday 30 November, and agree a check twice a day: store orders against purchases in each platform.
| Platform | Where to look | How fresh | Watch for |
|---|---|---|---|
| Store admin | Orders | live | the count everything is checked against |
| GA4 | Realtime, purchases | a few minutes | purchases stop while orders continue |
| Google Ads | Conversions (by conv. time) | hours behind | flat while orders rise |
| Meta | Events Manager, Purchase | last received time | volume falls, new diagnostics |
Two timing traps catch people on the Friday evening.
- GA4’s daily reports aren’t final on the day. Google says processing can take 24 to 48 hours and the numbers may change in that time. Use Realtime for “is it still firing”, and judge totals on Monday. If GA4 is still ahead of the store’s orders then, suspect a duplicate (F-101 ) before anything else.
- Google Ads dates conversions by the click, not the sale. A purchase on Friday from a click on Wednesday shows under Wednesday in the standard columns. The “Conversions (by conv. time)” columns date it on Friday, which is the view to put beside the day’s orders.
What if tracking breaks during the sale?
The common causes are dull: a late GTM publish, or an app update nobody mentioned.
- Roll back. Publish the GTM version you wrote down. If the cause is an app, switch off its tracking or ask the vendor.
- Write down when it started and ended. You’ll need both for the next step and for the post-sale report.
- Add a data exclusion in Google Ads for that window. It tells Smart Bidding (Google’s automated bidding) to ignore those hours and leaves reporting alone. Google says use it for tracking outages, not often and not for long; a sale that simply performed differently doesn’t qualify.
- Meta has no equivalent step. Record the window so the post-sale numbers can be read with it in mind.
- Re-test with one order before calling it fixed.
What should wait until December?
The rebuilds from the “doesn’t fit” list, any app you mean to remove that isn’t causing a fault, and the clean-up of old conversion actions. December is quieter, and a change made then can be watched across normal weeks before the January sales. If you want the whole setup checked rather than the fortnight’s list, that’s a tracking audit.
Common questions
What tracking should I check before Black Friday?
Place one real test order per payment method. Check it arrives once in GA4 and Meta with the real order number and the right value, and that the Google Ads conversion tag fires with the same ID and value. Check the sale keeps its source after the payment provider and the consent default is set before any tag fires. Fix what fails, then freeze changes.
What should I fix first if I find several tracking faults?
Fix them in the order they distort bidding: purchases missing or counted twice, then the wrong purchase value, then consent wired wrong, then the sale’s source lost at the payment provider. Google Ads and Meta bid on the purchases they’re sent, so faults in that signal cost money every hour of the sale. Naming and tidying wait.
When should I freeze tracking changes before Black Friday?
The Monday of sale week, 23 November 2026, at the latest. After that, no Google Tag Manager publishes, new apps, theme releases or conversion setting changes unless they fix a fault found in testing. Write down the live GTM version number first, so rolling back takes a minute.
Should I change my Google Ads conversion goals before Black Friday?
Not unless the current goal is wrong. Google says a significant change to the conversion goals a bid strategy optimises puts it into Learning, which never lasts more than 7 days. Budget and target changes don’t trigger it. Bidding and seasonality adjustments are your PPC lead’s call; the tracking job is to make sure the conversion they bid on is counted once, with the right value.
What if conversion tracking breaks during Black Friday?
Roll Google Tag Manager back to the last working version, note when the outage started and ended, and add a data exclusion in Google Ads for that window. It tells Smart Bidding (Google’s automated bidding) to ignore the conversion data from those hours and doesn’t change your reports. Meta has no equivalent step, so record the window for the post-sale report.
Why don’t GA4 and Google Ads match the store’s orders on Black Friday itself?
Timing first. Google says GA4 data processing can take 24 to 48 hours, and Google Ads reports conversions against the day of the ad click unless you add the “by conv. time” columns. Compare same-day orders with GA4 Realtime and the Ads “by conv. time” columns, and judge totals once the data has settled. If GA4 still runs ahead of the store, suspect a duplicate purchase.
Sources
Read on 11 November 2026:
- Google Analytics Help, minimise duplicate key events with transaction IDs (GA4 removes repeat purchases with the same transaction ID)
- Google Ads Help, use a transaction ID to minimise duplicate conversions (duplicates within the same conversion action)
- Meta for Developers, deduplicate Pixel and server events (matching
event_idand event name, same pixel, within 48 hours) - Google Analytics Help, DebugView (shows events from devices in debug mode, including via Tag Assistant or GTM Preview)
- Google Tag Manager Help, container versions (publish a previous version from Versions; publish history)
- Google Ads Help, bid strategy statuses (Learning after goal changes, never more than 7 days; budget and target changes don’t trigger it)
- Google Ads Help, seasonality adjustments (short events of 1 to 7 days)
- Google Ads Help, about data exclusions and use data exclusions for conversion data outages (tagging issues, website outages, data import issues; Search, Display, Shopping and Performance Max; reporting unaffected; not frequently or for long periods)
- Google Ads Help, conversions by time columns (standard columns by click date; “by conv. time” by conversion date)
- Google Analytics Help, data freshness (processing can take 24 to 48 hours; Realtime typically a few minutes)
- Google Analytics Help, consent mode modelling (behavioural modelling eligibility thresholds)
- Shopify Help Centre, upgrading the Thank you and Order status pages (Additional scripts and non-Plus deadline of 26 August 2026)