Google Ads Optimization: Beyond Platform-Reported ROAS

Google's reporting makes every campaign look like a winner

Google Ads has a structural incentive to make your campaigns appear profitable: the better your reported ROAS, the more you spend. This doesn't mean Google manipulates data, but the default reporting setup presents numbers in the most favorable light.

Three specific mechanisms inflate Google Ads ROAS beyond what your campaigns actually deliver:

1. Branded search mixed into account-level metrics
2. Performance Max's opaque multi-channel claiming
3. Broad match query expansion into low-intent terms

Understanding these mechanisms lets you optimize toward real performance rather than reported performance.

Mechanism 1: Branded search inflation

The problem

Branded search campaigns (bidding on your company name) produce 10-20x ROAS in Google's reporting. This looks spectacular but is mostly an artifact: people searching for your brand name are already your customers or already decided to buy. The ad intercepts a click that would have gone to your organic listing.

When branded search is combined with non-branded campaigns in account-level reporting, it inflates the aggregate numbers. An account spending $50K on branded (15x ROAS) and $150K on non-branded (2x ROAS) reports a blended 5.25x ROAS. The true productive ROAS from demand-generating campaigns is 2x.

The fix

Separate branded and non-branded reporting permanently. Never evaluate account performance on blended numbers. Create separate reporting views for:

• Branded exact match
• Branded broad match
• Non-branded search
• Shopping
• Performance Max
• Display
• YouTube

Run a branded search incrementality test. Pause branded ads in one geographic region for 3-4 weeks. Track organic click-through rates in the holdout region. Most brands find that 70-90% of branded clicks transfer to organic results, meaning 70-90% of branded search spend is wasted.

Reduce branded bids, don't eliminate them. If competitors bid on your brand terms, maintain a presence but reduce bids to the minimum needed to hold top position. If no competitors bid on your terms, consider pausing branded ads entirely.

Mechanism 2: Performance Max opacity

The problem

Performance Max (PMax) campaigns run across Search, Shopping, Display, YouTube, Discover, and Gmail simultaneously. Google reports aggregate performance but provides limited visibility into which channels within PMax drive results.

The issue: PMax frequently runs branded search queries within its campaign. When a user searches for your brand name and converts, PMax claims credit -- even though the conversion would have happened through your standard branded search campaign or organic listing.

Multiple analyses have found that 30-60% of PMax conversions come from branded search queries and remarketing audiences. The "smart" algorithm isn't finding new customers -- it's capturing existing demand and presenting it as PMax performance.

The fix

Add brand exclusions. Google now allows brand exclusions in PMax campaigns. Add your brand name and common misspellings to the exclusion list. This forces PMax to find conversions through non-branded channels.

Compare PMax to your previous campaign structure. Before PMax, you had separate Shopping, Search, Display, and YouTube campaigns. If PMax is producing higher ROAS but lower total conversions than your previous structure, it's likely concentrating spend on the easiest conversions (branded, remarketing) and neglecting prospecting.

Review the PMax asset groups report. Although limited, Google provides some breakdown of PMax performance by asset group. Look for asset groups with high conversion volumes but suspiciously low CPAs -- these are often running on branded queries.

Run a PMax incrementality test. Pause PMax in select geographic regions and monitor total account performance. If conversions don't drop meaningfully in holdout regions (because other campaigns capture the same demand), PMax's incremental contribution is lower than reported.

Mechanism 3: Broad match query expansion

The problem

Google's broad match has expanded significantly in recent years. A broad match keyword like "running shoes" might trigger ads for queries like "best athletic footwear," "Nike sneakers on sale," or "marathon training gear." Some of these expansions reach genuinely new audiences. Others waste spend on irrelevant queries.

The problem for ROAS: broad match queries that convert tend to be close to your core terms (and thus would have been captured by phrase or exact match anyway). The expansions that reach genuinely new queries often have much lower conversion rates. But Google's reporting blends everything together.

The fix

Review search query reports aggressively. Check the actual queries triggering your broad match keywords weekly. Add irrelevant queries as negatives immediately.

Segment reporting by match type. Compare ROAS for exact match, phrase match, and broad match separately. If broad match ROAS is more than 40% lower than exact match ROAS, the expansions are diluting performance.

Use broad match with smart bidding, not with manual bids. Google's broad match works best when paired with Target CPA or Target ROAS bidding because the algorithm can bid low on long-shot queries and high on likely converters. With manual bids, you pay the same bid for every query regardless of intent.

Track incremental queries from broad match. The genuine value of broad match is reaching queries you wouldn't have thought to target. Identify queries from broad match that led to conversions and that aren't covered by your exact/phrase campaigns. If these incremental queries produce profitable conversions, broad match is adding value. If most converting queries are close variants of your existing keywords, broad match isn't finding new demand.

Building a real Google Ads performance picture

Step 1: Create a deducted ROAS calculation

Start with Google's reported numbers and subtract the inflated components:

1. Total Google Ads revenue (reported): $500K
2. Minus branded search revenue: -$150K
3. Minus PMax branded query revenue (estimated at 40% of PMax): -$60K
4. Non-branded productive revenue: $290K

Non-branded productive ROAS = $290K / (Total spend - branded spend) = your real number.

Step 2: Match to server-side conversions

Compare Google Ads reported conversions to actual verified conversions in your CRM or ecommerce platform. The difference includes:

• Cross-device modeled conversions (Google's estimate for users who clicked on one device and converted on another)
• Enhanced conversions (modeled matches using hashed email data)
• View-through conversions from Display and YouTube

Track the match rate over time. For most accounts, 55-75% of Google-reported conversions can be verified server-side. The remaining 25-45% are modeled.

Step 3: Build campaign-level true ROAS

For each campaign type, calculate both reported and server-side ROAS:

| Campaign Type | Reported ROAS | Server-Side ROAS | Inflation Factor | |--------------|---------------|------------------|------------------| | Branded Search | 15.2x | 1.8x (incremental) | 8.4x | | Non-Brand Search (Exact) | 3.8x | 2.9x | 1.3x | | Non-Brand Search (Broad) | 2.4x | 1.6x | 1.5x | | Shopping (Standard) | 4.2x | 3.1x | 1.4x | | PMax (brand excluded) | 2.8x | 1.9x | 1.5x | | PMax (no exclusions) | 5.6x | 2.1x | 2.7x | | Display | 0.8x | 0.3x | 2.7x | | YouTube | 1.2x | 0.9x | 1.3x |

This table immediately reveals where budget is productive and where it's inflated.

Optimization priorities based on real data

Priority 1: Reduce branded search waste

If your incrementality test shows branded search is less than 20% incremental, cut bids by 50% and monitor organic click share. Most brands can save 40-70% of branded search budget with less than a 5% impact on total traffic.

Priority 2: Clean up PMax

Add brand exclusions. Review search query insights for non-incremental queries. If PMax performance drops significantly after brand exclusion, the campaign was mostly capturing branded demand -- not generating new demand.

Priority 3: Shift budget to highest real-ROAS campaigns

Non-branded exact match search and standard Shopping typically show the smallest gap between reported and real ROAS. These campaigns tend to have the highest incremental value because they reach users actively searching for your product category.

Priority 4: Evaluate Display and YouTube on incremental metrics

Display and YouTube often look unprofitable on last-click ROAS but may drive incremental awareness that shows up through other channels. Test with geographic holdouts to measure the full cross-channel impact before cutting.

Frequently Asked Questions

Should I stop using Performance Max entirely?

Not necessarily, but you should run it with brand exclusions enabled and compare its real performance to a manual campaign structure. For some accounts, PMax with exclusions genuinely outperforms manual campaigns because Google's algorithm can efficiently allocate across Shopping, Display, and YouTube. For others, the opacity isn't worth the trade-off. Run a 4-week split test: PMax with exclusions in some markets, manual campaigns in others. Compare total revenue, not just PMax-reported conversions, to determine which structure produces better real results.

How much of my Google Ads budget is actually wasted on branded search?

For the typical ecommerce brand with reasonable organic SEO, 60-80% of branded search spend captures clicks that would have gone to organic results. The exact percentage depends on two factors: whether competitors bid on your brand terms (if so, branded ads have higher incrementality) and how prominent your organic listing is (if your site ranks #1 organically with sitelinks, branded ads add less value). An incrementality test gives you the precise number for your brand. Until you run one, a conservative assumption is that 50% of branded search spend is non-incremental.

Does Google's Smart Bidding compensate for the ROAS inflation?

Smart Bidding (Target CPA, Target ROAS) optimizes toward the conversion data it receives, including the inflated numbers. If you set a 4x Target ROAS and 30% of conversions are inflated, Google will optimize to a real ROAS of roughly 2.8x while reporting 4x. The algorithm is efficient at hitting its target -- but the target is based on inflated measurement. To compensate, set your Target ROAS higher than your actual target by the estimated inflation factor. If you want a real 3x ROAS and your inflation factor is 1.4x, set Target ROAS at 4.2x. This forces the algorithm to reach higher-quality audiences to hit the stricter target.

---

Go Funnel uses server-side tracking and multi-touch attribution to show you which ads actually drive revenue. Book a call to see your real numbers.