When Smart Bidding Breaks: A SaaS Google Ads Audit Case Study

Introduction

This case study comes from a Google Ads audit for a growth-focused advertiser selling a software product.

They were doing many things “right” on paper: campaigns were active, conversion tracking existed, and there were pockets of decent acquisition efficiency.

But the lived experience was different. Spend was increasing, results were inconsistent week to week, and the team could not confidently answer a basic question: “Are we paying more because the market got harder, or because our setup is leaking money?”

Here were the symptoms we saw in the last 30 days of data:

• Several campaigns had acceptable CPA, but others had meaningful spend with no business outcome
• The account CPA was dramatically higher than the target CPA on file
• Smart bidding strategies were everywhere, but conversion volume per campaign was thin
• Measurement looked “complete” at first glance, yet attribution signals hinted at double counting

Key takeaway: When an account is labeled “unstable,” it is rarely one big problem. It is usually a stack of small, compounding setup issues.

The situation (what we evaluated)

We audited the last 30 days with a purchase focused lens.

Even in SaaS, “purchase” can mean different things (paid subscription, first invoice, self serve checkout, or a post trial upgrade). For this audit, the account’s stated goal category was purchase oriented, so we evaluated everything against goal aligned purchase conversions.

A few campaign level observations set the tone:

• Some campaigns produced the best observed CPAs in the account
• Two campaigns had meaningful spend with zero recorded outcomes
• The account was flagged for significant waste signals overall

This is a very common SaaS pattern: the “good” campaigns keep the account alive, while the “noisy” campaigns quietly consume budget and distort learning.

What the audit revealed (diagnosis process)

We used a simple order of operations to diagnose what was really happening:

1. Confirm measurement and de duplication (are we counting the right thing once)
2. Validate smart bidding prerequisites (enough goal aligned conversions for the bid strategy)
3. Identify waste pockets (spend with no outcomes, poor query intent, weak landing alignment)
4. Prioritize fixes by what will stabilize learning fastest

What we found was not exotic. It was a set of highly fixable fundamentals.

1. Smart bidding was over deployed relative to conversion volume

A large share of campaigns were using automated bid strategies that require stable, goal aligned conversion volume to work well:

• Multiple campaigns were running Target CPA with too few goal aligned conversions
• Multiple campaigns were running Maximize Conversions with too few goal aligned conversions
• Multiple campaigns were running Maximize Conversion Value with too few goal aligned conversions

This matters because smart bidding does not optimize “for your business.” It optimizes for the conversion signals you feed it, at the volume you can sustain.

When volume is thin, Google’s model leans heavily on sparse signals, broader matching, and shaky assumptions. The result is often:

• Volatile CPAs
• Over expansion into marginal queries and placements
• Budget getting pulled toward whatever looks like a conversion, even if quality is mixed

Key takeaway: Smart bidding is not a cheat code. It is a multiplier. If the signals are thin or messy, it multiplies the mess.

The measurement trap: double counting “purchase” conversions

The audit flagged a critical measurement issue: the same purchase could be counted twice due to multiple sources being imported and treated as primary.

In this account, purchase conversions were effectively being reported from more than one pipeline (for example, app or analytics sources alongside platform native tracking).

Even when tracking is “working,” double counting breaks two things at once:

• Reporting credibility (you think you have more purchases than you do)
• Smart bidding learning (the system optimizes toward inflated or duplicated events)

The recommended fix was straightforward: keep one primary purchase source and set the rest to secondary.

For SaaS founders, the important nuance is this:

• You can keep multiple sources for troubleshooting and triangulation
• But you should only let one source steer bidding and be treated as the primary truth for optimization

Why CPA drifted so far from the target

The audit detected that actual CPA was far above the target CPA saved in the account.

This gap typically happens for one (or more) of these reasons:

• The target CPA was set during an earlier, easier period and never recalibrated
• Conversion quality degraded (more low intent traffic, worse landing experience, weaker lead to paid pipeline)
• Smart bidding chased “convertible” users but not necessarily ideal customers
• Duplicate conversions inflated conversion counts, masking true cost per real purchase

It also flagged that multiple campaigns were above the saved CPA target for purchase.

This is where teams often make the wrong move: they cut budgets broadly.

A better approach is to shrink the waste first, so you can keep the parts that are working while restoring stability.

Waste pockets: spend with no outcomes

Two campaigns were highlighted with meaningful spend and zero recorded outcomes.

In a SaaS account, this is almost never “bad luck.” It usually indicates at least one of the following:

• Intent mismatch (queries or audiences are not aligned to buyers)
• Tracking mismatch (the campaign drives conversions, but they are not attributed due to setup)
• Landing page mismatch (clicks are real, but the page does not fulfill the promise)
• Offer mismatch (too much friction, unclear pricing, weak trial proposition)

In audits, these campaigns become the best leverage point because you can often reclaim a material portion of budget quickly.

What we typically do next is isolate which failure mode is present:

1. Check search terms and placement quality (are we buying buyer intent)
2. Check conversion paths and attribution windows (are we failing to measure)
3. Check landing message match (does ad promise equal landing reality)
4. Check funnel friction (speed, form steps, trial flow, paywall surprises)

Quality Score below 7: the “invisible tax” on SaaS CPCs

The audit flagged multiple active keywords with Quality Score below 7.

Quality Score is not a vanity metric, but it is a very practical one: low QS often means you pay more per click for the same position.

For SaaS founders, the root cause is usually not “Google hates us.” It is misalignment across three layers:

• Keyword intent: what the user actually wants
• Ad relevance: whether your RSA headlines mirror that intent
• Landing page experience: whether the page confirms the promise quickly

Typical fix patterns that work:

• Break up ad groups so each one maps to a tighter intent cluster
• Write RSAs that repeat the language users type (not just brand slogans)
• Build intent specific landing sections (use cases, integrations, pricing clarity)

Key takeaway: The fastest QS improvements usually come from message match, not from design overhauls.

Creative and asset coverage: the app campaign gap

The audit also flagged that app campaigns were missing core creative asset coverage.

Even if you are a SaaS company, app campaigns behave like creative first systems. Thin asset sets constrain the algorithm and can lead to:

• Limited inventory access
• Over reliance on a small set of creatives that fatigue quickly
• Poorer ability to communicate the install value and the downstream in app action

A pragmatic asset plan looks like this:

• Add more visual variety (product UI, outcomes, social proof)
• Include at least one video or motion asset that communicates the install value fast
• Align creative to one clear promise (not five features at once)

The fix plan we recommended (stabilize first, then scale)

Rather than “optimize everything,” we recommended a stabilization sequence.

Step 1: Clean measurement so bidding learns from one truth

• Choose one purchase conversion source as Primary
• Set the other purchase sources to Secondary
• Verify that the primary purchase event matches the real business definition of paid conversion

Step 2: Simplify bidding until conversion volume supports automation

If a campaign does not have enough goal aligned conversions per month, do not force advanced smart bidding.

Common stabilization options:

• Use a simpler bid strategy temporarily
• Consolidate fragmented campaigns so conversions concentrate
• Tighten targeting and query intent so conversions become denser

Step 3: Cut or quarantine zero outcome spend

• Pause, cap, or isolate campaigns that spend without producing goal aligned conversions
• Audit search terms, placements, and audience expansion settings
• Rebuild with stricter intent controls and clearer landing alignment

Step 4: Lift Quality Score where it impacts CPC and CPA most

• Rebuild ad groups around tighter intent
• Rewrite RSAs to match the query language
• Update landing pages for faster confirmation (headline, proof, next step)

Lessons for SaaS founders running Google Ads

This audit reinforces a set of patterns we see repeatedly in SaaS accounts that feel “expensive” or “unstable.”

• Smart bidding is only as good as your conversion signal and volume
• Double counted conversions do not just break reporting, they break bidding
• Zero outcome spend is often a configuration problem, not a market reality
• Quality Score issues quietly raise your CPC floor
• Stabilization is a phase. Scaling comes after the account has clean signals and concentrated volume

If your Ads account feels like it is working and not working at the same time, that is usually a sign you have a few strong pockets plus a few silent leaks.

Want a second set of eyes?

If you want to pressure test your Google Ads setup, you can request an audit or try the VEOtool beta to scan for smart bidding readiness, measurement conflicts, waste pockets, and prioritised fixes.

No hype, just a practitioner style checklist of what is likely holding your account back and what to fix first.