The “Smart Bidding Is Broken” Audit: When 43 Primary Conversions Tell Google to Optimize for Consent Screens

Introduction

I see this pattern in audits more often than most teams want to admit: the account structure looks fine, creative isn’t the issue, targeting isn’t the issue… but Smart Bidding still behaves like it’s guessing.

In this anonymized case study, I’m breaking down a Google Ads audit for a growth-focused advertiser (think: high-intent product/service, aggressive acquisition goals). The audit window was the last 12 months, and our tooling flagged a “broken” growth score with significant waste signals.

The most important part: this wasn’t a “Google Ads is bad” problem. It was a measurement problem that quietly poisoned bidding.

Symptoms we saw (and you’ve probably seen too):

• Plenty of reported conversions, but they didn’t map to revenue or qualified leads
• Smart Bidding learning never seemed to stabilize (or it “stabilized” in a bad place)
• Hard to explain performance changes because attribution was effectively unusable

If Google’s optimizing for low-intent events, your best campaigns will still look like they’re underperforming.

The context: “Nothing is obviously wrong”… until you look at conversions

At a glance, the account looked healthy in a few ways:

• Campaign structure scored well (no major red flags)
• Audience/targeting setup wasn’t the culprit
• Creative and landing pages weren’t being flagged as primary issues

That’s why this kind of audit is dangerous: it passes the “smell test.” If you’re an agency owner inheriting this account, you can waste weeks testing ads and keywords while the real issue sits in conversion settings.

The audit surfaced:

• Critical issues: 9
• High issues: 26
• Medium issues: 7

The highest concentration of problems lived in:

• Conversion Tracking Integrity (multiple major issues)
• Bidding & Smart Bidding Health (major issues)
• Attribution & Measurement (effectively non-functional)

Diagnosis process: how we confirmed Smart Bidding was optimizing for the wrong actions

When we suspect conversion contamination, we don’t start by changing bids. We start by asking one question:

“What actions are set as Primary, and do they represent business outcomes?”

Here’s the quick diagnostic flow we used (and what I’d recommend your team standardize):

1. Pull the conversion action list and filter to “Primary”
2. Identify which are true outcomes (purchase, qualified lead, subscription, booked call) vs. micro-events (page_view, session_start, screen_view)
3. Check whether the goal-aligned conversion has recorded data in the audit window
4. Count the number of active primary conversions (anything above ~3 is a smell)
5. Validate attribution configuration: are we even able to trust the reported conversions?

What we found was extreme, but not rare.

Root cause #1: Page views and app engagement events were set as Primary conversions

Multiple PAGE_VIEW-like events were being treated as primary conversion actions.

Examples (generalized):

• Consent screen views (e.g., “medical_consent”) tracked as conversions
• App “screen_view” events tracked as conversions
• “session_start” tracked as a conversion

This is the classic trap: these events fire easily and frequently, so the account “shows conversions.” But they’re weak intent. They don’t represent a user completing the core objective.

Why this breaks performance:

• Smart Bidding learns faster from high-volume signals
• If the easiest “conversion” is a consent screen or a screen view, the system will optimize for users most likely to produce… more consent screens and screen views
• Your CPCs and targeting drift toward cheap behavior, not valuable outcomes

Smart Bidding is only as smart as the conversion definition you feed it.

Root cause #2: 43 Primary conversions active (recommended max: ~3)

The audit flagged 43 active primary conversions.

Even if a few of those were legitimate, the model is forced to average across a messy basket of actions. In practice, that means:

• “Easy” conversions overpower “hard” conversions
• Reporting becomes a vanity metric factory
• Bid strategies chase volume instead of value

For agency owners, this is where you get painful client conversations:

• “Why did conversions go up but revenue didn’t?”
• “Why did CPA improve but lead quality tank?”
• “Why can’t we scale this if ROAS is ‘good’?”

Because the account is measuring the wrong thing.

Root cause #3: No goal-aligned conversions detected for the past year

This is the part that really matters.

The audit detected no goal-aligned conversions in the last 12 months. In plain terms: the platform had no consistent record of the outcome the business actually cares about.

This can happen in a few ways:

• The true conversion action exists, but it’s set to Secondary
• The tag/event isn’t firing due to app/web implementation issues
• The wrong property/stream is connected
• A new site/app release changed event names and nothing was updated

Whatever the reason, the effect is the same:

• You can’t evaluate Smart Bidding quality against the real goal
• You can’t run clean experiments
• You can’t attribute improvements to strategy vs. tracking noise

The fixes: how we’d clean this up without tanking delivery

This is where agencies sometimes overcorrect. If you rip out conversions overnight, you can shock the system.

Here’s the safer, practitioner approach.

Step 1: Redefine “Primary” to only include real outcomes

The immediate action:

• Move all micro-events to Secondary
• Keep only 1–3 true business outcomes as Primary

A practical “Primary conversion” shortlist might look like:

• Purchase (if e-commerce)
• Qualified lead (form submit with qualification, booked call, or verified MQL)
• Subscription start (if that’s the monetization event)

Keep these as Secondary (helpful, but not bid-driving):

• page_view / screen_view
• session_start
• consent events
• add_to_cart (sometimes), view_item, scroll

Secondary conversions are not “useless.” They’re just not what you want Google to bid toward.

Step 2: Confirm the goal conversion fires reliably (then backfill learning carefully)

Before you change bid strategies or budgets, validate:

• The goal conversion triggers on the correct page/screen
• Duplicates aren’t firing
• Cross-domain/app handoffs don’t break the journey
• The conversion is mapped to the correct goal in Google Ads

Then, give Smart Bidding something real to learn from:

1. Start with a short stabilization window after conversion changes
2. Keep budgets steady while learning re-anchors to the new signal
3. Watch conversion volume and lag (especially if it’s a longer sales cycle)

If volume is too low, you have options:

• Use a higher-funnel but still meaningful proxy (e.g., “qualified lead step 2”)
• Import offline conversions (if lead gen)
• Improve tracking coverage (server-side, enhanced conversions, CRM integration)

Step 3: Rebuild attribution and measurement so performance is explainable

The audit’s measurement module was effectively at zero, with many high-priority issues.

Even if you don’t overhaul everything at once, agencies should aim for:

• Clean conversion taxonomy (naming, primary/secondary rules)
• Clear ownership of tags/events (who updates when site/app changes?)
• A reporting view that separates:
• Primary outcomes (what the business buys)
• Secondary signals (what helps diagnose)
• Engagement (what’s interesting but not decision-driving)

This is where you stop “arguing with the dashboard” and start making decisions.

Lessons for agency owners: how to spot this in the first 15 minutes

If you want a fast gut-check during onboarding, look for these tells:

• The conversion list includes page_view, screen_view, session_start as Primary
• There are more than 10 Primary conversions (let alone 40+)
• Reported conversions feel “too easy” relative to revenue/lead volume
• Smart Bidding performance swings without a clear cause
• The account can’t answer: “Which conversion equals money?”

A simple onboarding script that saves weeks:

1. “Show me your primary conversions.”
2. “Which of these directly produces revenue?”
3. “Has that conversion fired consistently in the past 30 days?”

If the answers are fuzzy, pause optimization and fix measurement first.

Closing: the unsexy fix that unlocks everything else

This case study is a good reminder that campaign structure can be perfect and performance can still be capped.

When conversion tracking integrity is broken, Smart Bidding doesn’t fail loudly—it fails quietly. You get activity without outcomes, and you spend time optimizing the wrong levers.

If you’re an agency owner, this is also a positioning advantage: teams that can diagnose conversion integrity quickly win retainers, because they stop the waste at the source.

Don’t scale spend until you trust the conversion signal.

If you want, you can try the VEOtool beta to scan an account for issues like “micro-events set as Primary,” bloated conversion lists, and measurement gaps—or request an anonymized audit walkthrough and I’ll share how I’d prioritize fixes in your situation.