Broken Conversion Tracking Is Wrecking Your Smart Bidding: 7 Silent Errors to Audit
When a Target CPA or Target ROAS campaign underperforms, the instinct is to blame the bidding strategy, tighten the target, or switch approach entirely. In most accounts we audit, the actual cause sits one layer down: the bidding algorithm is optimizing perfectly toward conversion data that is quietly wrong. Here are the seven errors that corrupt it most often, and none of them throw an obvious error message.
- Smart Bidding optimizes ruthlessly toward whatever conversion data it's given, accurate or not; a data problem looks identical to a strategy problem from inside the platform.
- None of the seven errors below trigger an obvious platform warning; they require a deliberate audit against an independent source of truth to catch.
- Website redesigns, checkout changes, and consent platform updates are the most common silent triggers for a tracking break.
- Fix the data before touching the bid strategy; changing targets against corrupted data just optimizes toward the wrong number faster.
Why bad data breaks Smart Bidding worse than bad strategy
Target CPA and Target ROAS bidding work by feeding conversion signals into a machine learning system that adjusts bids in real time to hit a stated goal. The system has no independent way to verify that a "conversion" it's counting represents a real customer, a real sale, or a real lead; it trusts the data it's given completely and optimizes toward it with total consistency. That consistency is what makes broken tracking so damaging: a manually managed campaign with a data problem produces inconsistent, obviously odd results a human might catch. A Smart Bidding campaign with a data problem produces confident, consistent, entirely wrong decisions, because the algorithm is doing what it was built to do with the signal it received.
This reflects a broader shift in how Google Ads accounts are run in 2026: platforms are making more automated decisions, and the practical form of "control" that remains isn't manual bid adjustment, it's making sure the data feeding the automation is accurate. Everything below assumes your account already has conversion tracking configured; the point is that "configured" and "accurate" are not the same thing.
1. Duplicate conversion firing
The same conversion event fires more than once per user action
Common causes: a conversion tag placed on both a "thank you" page and a confirmation modal that appears before the redirect, a page refresh re-triggering an already-fired tag, or a single form submission triggering both a native form-fill tag and a separate destination-URL-based tag. The bidding algorithm sees twice the real conversion volume at what looks like half the real cost per conversion, and optimizes toward acquiring more of what it believes is a cheap, abundant outcome.
2. Conversions counted on the wrong page
A conversion tag fires on a page reachable without completing the actual goal action
A frequent version: a "thank you" page URL that's also reachable by directly navigating to it, bookmarking it, or hitting back-then-forward in the browser after an earlier real conversion, each of which can re-fire a URL-based conversion tag without a new genuine action occurring.
3. Consent mode suppressing data silently
Consent Mode is misconfigured and quietly withholding conversion signals
Since Google's consent mode changes, covered in our consent mode and ad storage guide, a misconfigured consent banner or default-deny setup can suppress a meaningful share of real conversions from ever reaching Google Ads, without any visible error. The campaign appears to underperform when the actual conversions happened but were never reported.
4. Mismatched attribution windows
The conversion action's attribution window doesn't match the actual sales cycle
A 7-day click attribution window on a product with a 30-day typical consideration period systematically undercounts real conversions, since many genuine sales complete after the window closes and never get credited back to the original click.
5. Cross-domain tracking breaks
A click ID gets lost when a user moves between domains or subdomains mid-journey
Common in accounts where checkout lives on a different domain than the marketing site, or where a booking widget is hosted by a third-party platform. Without proper cross-domain linking, the click that brought the user in loses its ad-click ID by the time the conversion fires, and the conversion either goes unattributed or gets misattributed to a different source.
6. Incorrect or missing conversion values
Conversion value is hardcoded, missing, or doesn't reflect actual order value
Target ROAS bidding depends entirely on accurate value data. A hardcoded placeholder value left in from initial setup, a value that doesn't account for discounts or refunds, or a value field that silently fails to populate on certain checkout paths all feed the algorithm a distorted picture of which conversions are worth pursuing.
7. Stale tags after a site migration
A redesign or CMS migration changed page structure without updating tag trigger conditions
A GTM trigger built around a specific URL path, button ID, or page element that no longer exists after a redesign will silently stop firing, or worse, start firing on the wrong new element that happens to share an old selector. Sites rarely QA conversion tracking as part of a redesign checklist the way they QA visual design.
Running the audit
- 01
Establish a source of truth
Your CRM, order management system, or actual sales team records, not Google Ads' own reported numbers, which is what's being audited.
- 02
Compare conversion counts over an identical date range
A gap beyond normal attribution-window variance points directly at one or more of the seven errors above.
- 03
Use Tag Assistant or GTM preview mode to watch tags fire live
Walk through your own conversion path manually and confirm exactly one tag fires, on the right trigger, with the right value.
- 04
Fix the data before touching bid strategy
Adjusting a Target CPA or Target ROAS goal against corrupted data just re-anchors the algorithm to a new, still-wrong number. Clean the signal first.
This exact audit sits at the center of our analytics and attribution service, and pairs directly with the broader process in our tracking and technical audit checklist and conversion tracking guide.
Frequently Asked Questions
Compare your Google Ads conversion counts against a source of truth, your CRM or actual sales records, over the same date range. A gap of more than a few percentage points, duplicate conversions, or conversions firing on the wrong pages are all signs the bidding algorithm is optimizing toward inaccurate data rather than a strategy problem.
It can tolerate minor noise, but systematic errors, duplicate firing, wrong-value conversions, or tracking that misses a meaningful share of real conversions, compound over time because the algorithm keeps reinforcing decisions based on the flawed signal. The cleaner the underlying data, the faster and more accurately Smart Bidding converges on an efficient strategy.
At minimum quarterly, and immediately after any website redesign, CMS migration, checkout flow change, or consent management platform update, since these are the events most likely to silently break existing tracking without triggering an obvious error.
Sometimes, through diagnostics flags in the Conversions section, but these checks don't catch every failure mode, particularly duplicate counting, incorrect attribution windows, or conversions that fire but carry inaccurate values. A manual audit against an independent source of truth remains necessary.
Use Tag Assistant to confirm the conversion tag fires once, on the right action, then check the conversion status in Google Ads. Finally, compare the conversions reported with the leads or sales in your CRM.
Ready to fix what’s costing you conversions?
We’ll review your acquisition funnel, show you what’s underperforming, and hand you a clear, prioritised plan, whether or not you choose to work with us.
Get a clear diagnosis of your acquisition performance.
- 30-minute review
- Free, with no obligation
- Clear findings and priorities
- You keep the findings
30 minutes · Free · No obligation · You keep the findings