Is your conversion tracking lying to you?

Conversion tracking rarely lies loudly. It drifts quietly, and because Smart Bidding optimizes toward whatever you count, bad tracking doesn't just mislead reports. It steers spend.

The common failures

Failure	What it looks like	Effect
Silent conversion	An action that usually records daily shows zero for days	Bidding sees "no results" and pulls back, or reports look like a collapse
Double counting	Two primary actions for one purchase or lead, e.g. a Google Ads tag and a GA4 import	Conversions and ROAS inflated
Micro-conversion as primary	Add-to-cart or begin-checkout set as primary with cart value	ROAS several times higher than real
Missing value	A value-based bidding campaign with no conversion values	The bid strategy has nothing meaningful to optimize
Wrong currency or scale	Value sent in a different currency, or in cents	Values off by a fixed factor
Inherited goals	A campaign uses account-default goals that include micro-actions	That campaign optimizes toward the wrong action

Worked example: a micro-conversion counted as a sale

Worked exampleIllustrative example; the numbers are not real account data

An online store spends $10,000 a month on a Target ROAS campaign. Two actions are primary: Purchase, and Add to cart, which also sends the full cart value.

	Conversions	Conversion value	ROAS
Purchases (real)	400	$40,000	4.0
Add to cart (counted as primary)	1,600	$96,000	-
What Google Ads reports	2,000	$136,000	13.6

Example

ROAS in the report: 13.6
ROAS the business sees: 4.0

The campaign is learning to find people who add to cart, not people who buy. Moving Add to cart to secondary fixes the report and the signal. Expect a re-learning period, and expect reported ROAS to drop: that drop is the truth arriving, not performance getting worse.

The same pattern hits lead generation: a thank-you page view plus a form-submit event, both primary, counts every lead twice.

A tracking health check

List primary actions in Goals and confirm each represents one real business outcome.
Check each campaign's goals, not only account defaults.
Look for silence: compare each primary action's last 72 hours with its usual daily volume.
Look for duplicates: a tag plus a GA4 import, or two tags on one thank-you page.
Verify values against a week of actual orders, net of refunds and cancellations, in the same currency.
Turn on Enhanced Conversions where it fits, so conversions can still be matched without cookies.
Test after every site change with a real order or lead, and confirm the value in Google Ads.
Check the click ID survives your landing page redirects.

How often

Check	When
Silent primary actions	Hourly or daily, automated
Reported value vs. real orders	Weekly
Primary/secondary setup and campaign goals	Monthly, and after any change
End-to-end test conversion	After any site, checkout or app change

Even healthy tracking isn't your order book. Reconcile platform conversions against your own order or CRM data, and label unchecked numbers as exactly that.