Broken google ads conversion tracking setup is the most expensive invisible problem in ecommerce, because in the dashboard it looks identical to weak performance. The responses to those two problems are opposites. This is the diagnostic we run before trusting any number in any account.
Why a broken Google Ads conversion tracking setup looks like weak performance
A campaign reporting poor ROAS tells you one of two stories. Either the ads genuinely are not selling, or they are selling and the sales are not being counted. The dashboard renders both stories with the same red numbers.
The danger is that the correct responses point in opposite directions. Weak performance calls for cutting budgets, pausing products, tightening targets. Broken measurement calls for the opposite: fix the counting, because the campaigns may deserve more budget, not less.
Guess wrong and you compound the damage. We have watched stores throttle their best campaigns for months because a tag silently died, which is why tracking comes first in every Google Ads engagement we take on. Diagnosis before treatment, always. The five checks below run in roughly an hour and settle which story your dashboard is telling.
The 5-minute backend check: Shopify orders vs Google Ads conversions
The fastest sanity check needs no tools beyond two browser tabs. Open Shopify and count orders for the last 30 days. Open Google Ads and look at total conversions for the same window.
Google Ads should report fewer conversions than Shopify has orders, because plenty of orders come from email, organic, direct, and social. That gap is healthy. What flags a problem is the impossible: more conversions than total orders, which usually means duplicate firing, or a conversion count near zero while ads drive heavy traffic, which usually means the purchase tag is not firing at all.
Five minutes, and you already know whether the deeper audit is urgent. If the numbers look impossible in either direction, stop optimizing and start investigating. We keep a fuller version of this self-check at how to tell if your conversion tracking is broken.
GA4 vs Google Ads discrepancy: what's normal and what flags a problem
The two Google products will never agree, and that is by design. Google Ads claims a conversion for any purchase it assisted within its attribution window. GA4 distributes credit across the whole journey. Different rulebooks, different scores.
A moderate, stable gap between them is the normal cost of attribution. What should worry you is a gap that is enormous, or one that suddenly changes shape. If Google Ads reports several times the purchases GA4 sees from paid traffic, suspect double counting. If GA4 sees paid purchases that Google Ads barely registers, suspect a dead conversion tag or a broken account link.
The skill is not making the numbers match. It is knowing your account's normal gap, so you notice the week it stops being normal.
Consent Mode v2: the EU compliance gap that silently kills conversion data
If you sell into the EU or UK, Google requires Consent Mode v2 signals before it will use your conversion data properly. Without them, conversions from consented users can go uncounted and Google's modelling of unconsented users never activates. The data simply thins out.
The cruel part is the silence. Nothing turns red. No error email arrives. Your campaigns just start learning from a fraction of reality, Smart Bidding gets cautious, and the account drifts downward in a way that looks exactly like creative fatigue or rising competition.
The check: confirm your cookie banner is a Google-certified consent platform, that it transmits the consent state to your tags, and that consent signals show as active in the Google Ads diagnostics. Many Shopify setups have a banner that looks compliant and passes nothing.
GTM audit: the container changes that break tags without touching the tag itself
Most tracking deaths are not caused by someone editing the conversion tag. They are caused by something the tag depended on.
A theme update renames the element a trigger was listening for. An app injects its own data layer and overwrites the purchase event. A developer cleaning up old scripts removes the variable that fed the conversion value. A checkout migration changes the page the tag expected to fire on. In every case the tag itself sits untouched and innocent in the container, firing into the void.
The audit is to trace the chain end to end: trigger, variables, data layer event, firing page. Check the container's version history against the date your numbers went strange. The culprit is usually a change nobody connected to tracking at all.
How to test the purchase event without placing a real order
You do not need to buy your own products to verify the most important tag you own. GTM preview mode connects to your live store and shows every tag evaluating in real time as you browse, add to cart, and walk to the payment step.
For the final step, a 100% discount code or a manual test gateway lets you complete a zero-value checkout and watch the purchase event fire with your eyes on the debugger. Check the value, currency, and transaction ID arriving with it, since a tag that fires with empty parameters is only half alive.
Then confirm receipt at the destination: the conversion should appear in Google Ads diagnostics as recently recorded. Fire and receipt are two checks, not one.
When tracking is confirmed broken: the priority fix sequence
Fix in this order: the purchase event first, since nothing else matters while revenue is invisible. Consent Mode second, so the data you now collect is allowed to be used. Deduplication third, so you are not counting double. Value and currency accuracy last, so Smart Bidding optimizes toward real margins rather than ghost numbers.
Then wait before judging performance, because the algorithms need time to relearn from honest data. In a bathroom retailer account we manage, getting the measurement foundations right came before everything else, and the account went on to lift ROAS by 146%.
If you suspect your numbers but cannot prove it, that is exactly what our free 48-hour audit exists for. We check the tracking before we judge a single campaign.