Why Performance Max ROAS looks fake — attribution inflation, view-throughs, and what to trust instead

Performance Max ROAS often looks fake because the campaign can claim credit for conversions it barely caused — especially view-through (impression) conversions, brand searches you already owned, and soft primary actions that are not real sales. Treat in-platform PMax ROAS as a directional diagnostic. Judge the campaign with MER (total revenue ÷ total ad spend), CRM-qualified rate, and click-based outcomes before you scale budget.

This article owns the measurement truth problem. For setup, prerequisites, and kill criteria, use the Performance Max SMB playbook. For UK service lead quality (form spam, LSA overlap, OCI), see PMax lead quality UK. For US home services lead quality (call floors, CRM stages, aggregator junk), see PMax lead quality US. For sequencing PMax against keyword Search, see Performance Max vs Search. For YouTube / Demand Gen lead gen (same view-through caution outside PMax), see Demand Gen & YouTube for SMB lead gen. For tracking rebuilds before any bidding AI gets clean signal, use the conversion tracking setup guide.

What “attribution inflation” means on PMax

Attribution inflation is when the reported conversion value or count exceeds the incremental business result. On Performance Max it shows up in three common patterns we see in US and UK SMB audits:

1. View-through credit — someone saw a Display or YouTube impression, later converted on brand Search or direct, and PMax still gets counted depending on settings and columns.
2. Brand cannibalization — PMax spends on (or near) your brand queries and reports strong ROAS on demand branded Search already captured cheaply.
3. Soft conversion primaries — form starts, thank-you refreshes, or micro-events marked primary. PMax finds cheap volume; the dashboard celebrates; cash does not move.

None of these require Google to be “lying.” The platform optimizes and reports against the conversion goals and attribution settings you defined. Bad goals produce confident nonsense.

View-through conversions — what they are and why SMBs misread them

A view-through conversion typically means a user saw an ad impression (did not necessarily click), then converted within the view-through window. That path is more common on YouTube, Display, and Discover inventory inside Performance Max than on classic Search.

For awareness brands with long consideration cycles, view-through can be a useful secondary signal. For service SMBs optimizing to booked jobs, treating view-through-heavy ROAS as proof of paid efficiency is how accounts scale into waste.

Questions operators actually ask

• “Why is my PMax ROAS so high overnight?” Often a mix of view-through credit, branded overlap, and a conversion action that fires too early.
• “Should I use click-only conversions?” For lead-gen truth tests, yes — analyze click-based (and CRM) outcomes separately from view-through-inflated columns. Do not only stare at All conversions.
• “How much of PMax is brand?” Enough that brand exclusions and a funded branded Search campaign are mandatory controls — covered in the playbook; measured here via query/category insights and branded Search CPA vs PMax “efficiency.”

Click-only truth tests (how we audit)

When a client says “PMax is crushing it,” we run a structured before/after pattern — not a vibes review.

Step 1 — Split the columns

In Google Ads, compare for the PMax campaign and for the account:

• Conversions (primary, click-based where configured)
• All conversions (often includes view-through and secondary)
• Conversion value vs CRM pipeline value for the same dates

If All conversions vastly exceeds primary click conversions while CRM is flat, you are looking at inflation pressure.

Step 2 — Hold brand Search constant

Keep branded Search funded. Apply brand exclusions on PMax. Re-measure PMax CPA/ROAS after exclusions stabilize (typically 1–2 weeks). If PMax efficiency collapses after brand is protected, the previous ROAS was partly rented brand credit.

Step 3 — Align primary actions to money

Demote soft events. Keep one primary that matches qualified lead, booked job, purchase, or offline closed-won. Soft primaries make every AI campaign look smart — including PMax.

Step 4 — Import reality

Where sales close offline, push qualified or won stages back with offline conversion imports. Smart Bidding then stops optimizing purely to junk forms.

Step 5 — Score the account, not the narrative

Run Performance Max Readiness Score for a go/no-go before you scale. Pair it with Google Ads Health Score and review tracking purity with Tracking Trust Score. Inflated ROAS on a low trust score is a red flag, not a win.

MER vs ROAS — the metric that survives platform theatre

Platform ROAS = conversion value Google attributes to the campaign ÷ campaign spend (within Google’s model and settings).

MER (marketing efficiency ratio) = total business revenue (or gross profit, if you are disciplined) ÷ total ad spend across channels for the same calendar period.

Why MER matters for PMax debates:

• It does not care which campaign claimed the last click or view.
• If you raise PMax spend and MER does not improve (or worsens) while PMax ROAS looks elite, the campaign is likely reallocating credit, not creating demand.
• MER pairs with CRM: qualified leads, close rate, and capacity. A cheaper CPL that destroys close rate is not efficiency.

We expand the four-metric operating system in SMB marketing metrics that matter. For PMax specifically: never greenlight a budget increase on PMax ROAS alone.

Practical rule we use on engagements:

| Signal | Interpretation | Action | |--------|----------------|--------| | PMax ROAS ↑, MER flat, CRM flat | Credit reshuffle / inflation | Hold budget; fix goals, brand exclusions, click vs all columns | | PMax ROAS ↑, MER ↑, qualified rate stable | Likely real lift | Scale in 10–20% steps | | PMax CPL ↓, qualified rate ↓ sharply | Optimizing junk | Change primary conversion; pause scale | | PMax “wins” only with brand included | Cannibalization | Enforce brand exclusions; fund branded Search |

Brand cannibalization — the quiet ROAS factory

Performance Max can show for brand-related queries when Search is limited, exclusions are missing, or inventory overlaps. The campaign then reports strong returns on traffic that would have converted through branded Search at a lower CPA.

Symptoms:

• Branded Search volume or IS drops when PMax launches
• PMax search category insights skew toward brand and near-brand
• Account ROAS looks stable while incremental new-customer pipeline does not

Fixes (measurement + control): brand exclusions, adequate branded Search budget, negatives for irrelevant variants, and judging incrementality with MER + new-customer counts — not PMax ROAS in isolation.

Setup detail lives in the PMax playbook. Operator manual for exclusions, negatives, and the anti-cannibalization test: PMax brand exclusions & negatives. This insight’s job is to stop you from calling cannibalization “performance.”

Lead gen vs ecommerce — inflation looks different

Lead gen (home services, automotive workshops, local B2C): inflation usually means view-through + soft forms + untracked calls/WhatsApp. The dashboard overstates web-form efficiency while phone closes the business. Instrument calls and chats; import CRM stages.

Ecommerce: inflation often means view-through on Display/YouTube plus brand Shopping overlap. Use new-customer metrics, MER, and product-level contribution — not only campaign ROAS. Feed quality still matters; attribution theatre does not fix a broken catalog. Sequencing Shopping vs PMax: Shopping & Performance Max for product SMBs.

US and UK legal/consent stacks differ, but the diagnostic logic is the same: separate claimed credit from incremental cash.

Before / after audit pattern (what we write in scorecards)

When we deliver a free Google Ads audit and PMax is already live, the written scorecard usually includes:

1. Conversion inventory — every primary/secondary action, counting method, inclusion in “Conversions.”
2. Inflation score — gap between All conversions, primary click conversions, and CRM for 14–28 days.
3. Brand overlap check — exclusions present? Branded Search healthy?
4. MER snapshot — month-level spend vs revenue/pipeline (client-provided).
5. Decision — scale / hold / rebuild goals / pause PMax until tracking passes.

If you want that scorecard on your account: request a free Google Ads audit. Bring CRM export dates that match the Ads date range — without that, everyone argues about dashboards.

What to do this week (operator checklist)

1. Export PMax conversions vs All conversions vs CRM for the last 28 days.
2. Confirm brand exclusions are on; branded Search is funded.
3. Demote soft primaries; keep one money action.
4. Add MER to the weekly review next to platform ROAS.
5. Read the Performance Max SMB playbook only after steps 1–4 — setup on dirty data teaches the AI the wrong lesson faster.

FAQ — Performance Max attribution

Why is my Performance Max ROAS so high?

Usually a mix of view-through credit, brand overlap, soft conversion actions, or value settings that do not match cash. High ROAS is a hypothesis, not a bank deposit. Validate with MER and CRM.

What are view-through conversions in PMax?

Conversions credited after an ad impression (not necessarily a click) within the view-through window. More common on video and display inventory inside PMax. Useful as a secondary lens; dangerous as the sole success metric for lead-gen SMBs.

Should I use click-only conversions?

For diagnosing inflation and training bidding on lead quality, prioritize click-based and offline/CRM outcomes. Keep view-through visible for learning, but do not scale budget because All conversions look pretty.

What is MER vs ROAS?

ROAS is platform-attributed return for a campaign. MER is business revenue ÷ total ad spend. MER survives cross-channel credit fights; ROAS does not. Use both — MER to decide scale, ROAS to debug.

How much of Performance Max is brand cannibalization?

Enough that you must measure it. If excluding brand collapses PMax “efficiency,” a large share of prior ROAS was brand credit. Fund branded Search and keep exclusions on for most SMBs.

Does fixing attribution lower my reported ROAS?

Often yes — temporarily. That drop is usually honesty, not failure. After goals and exclusions are clean, expect the algorithm to recalibrate over 2–4 weeks. Judge the new baseline with MER and qualified pipeline.

Related frameworks

• Smart Bidding SMB playbook — Max Conv vs tCPA vs tROAS decision tree
• Performance Max SMB playbook — when to use PMax, setup, diagnostics, kill criteria
• Performance Max vs Search — decision matrix and SMB sequencing
• Google Ads conversion tracking setup — fix primaries before trusting any ROAS
• SMB marketing metrics that matter — MER, CAC payback, pipeline
• Advantage+ incrementality — Meta twin: why Advantage+ ROAS lies and how to test lift
• 47-point Google Ads audit checklist — structured account review
• Performance Max launch checklist — 38 pre-flight checks before spend
• Performance Max Readiness Score — Ready / Conditional / Not ready before launch
• Google Ads Health Score — five-minute account diagnostic

Free Google Ads audit — senior-led scorecard in 24 hours. We will tell you whether your PMax ROAS is signal or theatre.