When Smart Bidding Breaks: A SaaS Google Ads Audit

Introduction

This case study comes from a recent Google Ads audit for a growth focused advertiser (think SaaS with a purchase or paid conversion event).

On the surface, the account looked like it was “doing all the right things”: Smart Bidding everywhere, multiple conversion events configured, and lots of segmentation across campaigns.

But the audit score came back broken. Not because of one dramatic mistake, but because several small measurement and bidding choices stacked together until the system could not learn.

Symptoms we saw in the last 30 days:

• Multiple campaigns spending with zero goal aligned conversions
• Many segments (keywords, devices, locations) exceeding the saved CPA threshold without producing the real outcome
• Smart Bidding strategies applied in places where there was not enough conversion volume for them to work reliably

Key takeaway: when Google Ads is “active” but outcomes are missing, it is often a measurement and learning problem, not a creative problem.

The situation: busy account, low learning

The audit window was the last 30 days. Data confidence was moderate, and the waste signal was material. In other words, there was enough evidence to act, and the account had meaningful spend leaking into places that were not producing.

A few campaigns had acceptable CPAs relative to the target. That is normal. Most accounts have some pockets of efficiency.

The problem was that other campaigns had meaningful spend with no business outcome at all.

For SaaS founders, this pattern matters because it creates a false sense of traction. Dashboards show activity, impressions, clicks, even “conversions”, while pipeline and revenue stay flat.

How we diagnosed it (the audit process)

We did not start with ads or landing pages. We started with the two questions that determine whether Smart Bidding can help you or hurt you:

1. What exactly is Google optimizing for?
2. Is there enough clean signal for the algorithm to learn?

From there, we worked outward:

1. Conversion tracking integrity (are we counting the right thing, once?)
2. Bidding strategy health (does the bid strategy match volume and maturity?)
3. Audience and targeting quality (where is spend accumulating without outcomes?)
4. Attribution and measurement consistency (do sources agree, or double count?)
5. Only then: creative and landing page alignment

This sequencing is important. If measurement is noisy, Smart Bidding can “optimize” perfectly toward the wrong goal.

What the audit revealed (root causes)

1) Too many primary conversions diluted optimization

The account had 6 primary conversions active, when the recommended maximum is usually closer to 1 to 3 for most SaaS setups.

Why this breaks performance:

• Smart Bidding treats all primary conversions as success signals.
• If you mix “hard outcomes” (paid signups or purchases) with “easy outcomes” (micro events), the algorithm often finds the easiest path.
• You can end up paying to generate a high volume of low value actions that look good inside Google Ads but do not move revenue.

Practical SaaS example:

• “Purchase” and “Trial started” might both be valuable.
• “Pricing page view” and “Button click” are usually not something you want to bid around as primary goals.

Key takeaway: if everything is a primary conversion, nothing is.

2) Duplicate purchase counting across measurement sources

We found a classic issue: the same purchase could be counted twice because multiple sources were reporting it (for example, Firebase, GA4 import, and Google Ads tags).

This is more damaging than it sounds.

• Smart Bidding thinks conversion volume is higher than it really is.
• Reported CPA looks artificially low.
• The system may scale into segments that are not actually producing.

Founders often first notice this when:

• Google Ads shows more purchases than the backend.
• CPA looks “too good to be true”, but cash does not match.

Fix principle:

• Choose one primary purchase source.
• Set all other purchase imports to secondary for reporting only.

3) Smart Bidding applied without enough real conversions

A large portion of the account was running Smart Bidding strategies that require stable conversion volume:

• Multiple campaigns on Target CPA with too few goal aligned conversions
• Many campaigns on Maximize Conversions with too few goal aligned conversions
• At least one campaign on Target ROAS with too few goal aligned conversions

This is one of the most common SaaS scaling traps.

Smart Bidding is not magic. It is a system that needs enough clean examples of the desired outcome. A simple rule of thumb used in many accounts is 30+ real conversions per month per campaign (or conversion action set) before expecting stability.

If you do not have that volume:

• Bid strategies swing aggressively.
• Learning phases reset often.
• The system overreacts to noise and underweights true buyer intent.

4) Spend accumulating in “no outcome” segments (keywords, devices, locations)

Once measurement and bidding were misaligned, the waste pattern showed up clearly in the segments:

• 16 active keywords spent above the target CPA threshold with zero goal aligned conversions
• 25 device segments exceeded the CPA threshold with zero goal aligned conversions
• 24 location segments exceeded the CPA threshold with zero goal aligned conversions

This does not mean those segments are always bad.

It means the account had already paid enough in those areas to earn at least a few outcomes, and it did not. That is a strong signal to intervene.

Common underlying causes we see in SaaS:

• Broad or loose match types pulling in “research intent” queries
• Mobile traffic clicking but failing to convert due to friction (forms, speed, auth issues)
• Locations outside true market coverage still being targeted “just in case”

Key takeaway: segmentation is not just reporting. It is where you find spend that has already failed its trial.

What we fixed first (the highest leverage changes)

This is the order we recommend for founders and growth teams because it reduces risk.

Step 1: Clean up conversion priorities

1. Identify the single best indicator of revenue (purchase, paid signup, qualified lead).
2. Set that as Primary.
3. Keep only 1 to 2 supporting outcomes as Primary if truly necessary.
4. Move all micro events to Secondary.

This improves Smart Bidding behavior without changing budgets or creative.

Step 2: Remove duplicate counting

1. Audit all conversion actions that represent the same real world event.
2. Keep one canonical source as the “truth”.
3. Ensure the others are secondary or excluded.

If you are a SaaS founder, you want your Google Ads purchase counts to reconcile reasonably with backend billing or subscription events. Perfect parity is rare, but large gaps usually mean a setup issue.

Step 3: Match bid strategies to maturity

With low conversion volume, we typically move away from aggressive goal based automation until signal improves.

Practical options:

• If volume is low: consider simpler bidding (for example, manual control or less constrained automation) while you build conversion history.
• If volume is moderate but noisy: reduce constraints (avoid tight Target CPA or Target ROAS) and stabilize learning.
• If volume is healthy and tracking is clean: reintroduce Target CPA or Target ROAS with realistic targets.

The point is not to avoid automation forever. It is to earn the right to use it.

Step 4: Stop the bleeding in zero outcome segments

We used the CPA threshold as a triage mechanism. If a segment has spent enough to exceed your target and still has zero goal aligned conversions, it goes into a review queue.

Actions that usually help quickly:

• Tighten match types on the worst spending keywords
• Add negatives based on irrelevant query themes
• Apply device bid adjustments or split mobile into its own campaign if landing experience differs
• Constrain geo targeting to proven markets, or separate expansion geos into controlled tests

This is where you often reclaim a meaningful chunk of waste without changing top line strategy.

Lessons learned for SaaS founders

If you are running Google Ads for SaaS, these are the principles to take away.

• Optimize for one primary outcome you can defend financially.
• Do not let “easy conversions” steer bidding unless you are intentionally running upper funnel.
• Avoid strict Smart Bidding targets when conversion volume is low or tracking is messy.
• Use zero outcome spend as a decision rule. Past a threshold, segments must either improve or be paused.
• Reconcile your numbers. If Ads says you have far more purchases than billing, trust billing and fix measurement.

If your account feels “busy” but revenue is not moving, your system is probably optimizing toward the wrong scoreboard.

A simple checklist you can run this week

If you want a practical starting point, run this in your account:

1. Count your primary conversions. If it is more than 3, decide what to demote.
2. Check whether the same purchase is imported multiple ways.
3. For each Smart Bidding campaign, ask: do we have 30+ real goal conversions per month?
4. Pull keyword, device, and location reports and filter for segments with spend above your threshold and zero outcomes.
5. Make one change set at a time, then watch performance for stability.

Closing

This audit was not about finding one silver bullet. It was about restoring signal clarity so the account could learn again, then cutting waste in segments that had already proven unproductive.

If you are seeing similar symptoms, you do not need a complete rebuild. You need a clear primary conversion, consistent measurement, and bidding strategies that match your current volume.

If you want, you can try the VEOtool beta to scan for these patterns quickly, or request a lightweight audit and we will point out the highest leverage fixes first.