google-ads-pmax-search-cannibalization-detector

PMax Search Cannibalization Detector

Framework for detecting and resolving Performance Max stealing Search campaign traffic. The #1 frustration in the PPC community -- PMax claims branded conversions making ROAS appear artificially high.

Quick Diagnosis

IS PMAX CANNIBALIZING YOUR SEARCH?
====================================

Step 1: Check PMax search terms report
        -> Many branded terms? --> PROBABLY YES

Step 2: Compare blended vs non-brand ROAS
        -> PMax ROAS 2x higher than Search? --> RED FLAG

Step 3: Check Search campaign metrics since PMax launch
        -> Search impressions/clicks dropped? --> CONFIRMED

RISK LEVELS:
+-- LOW:      <20% brand traffic in PMax search terms
+-- MEDIUM:   20-50% brand traffic -> monitor + consider exclusion
+-- HIGH:     50-70% brand traffic -> enable brand exclusion
+-- CRITICAL: >70% brand traffic -> PMax ROAS is misleading

Step 1: Data Collection

1A: PMax Search Terms Report

Use google_ads_run_gaql to retrieve PMax search terms:

SELECT campaign.name, search_term_view.search_term,
    metrics.impressions, metrics.clicks, metrics.cost_micros,
    metrics.conversions, metrics.conversions_value
FROM search_term_view
WHERE campaign.advertising_channel_type = 'PERFORMANCE_MAX'
AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.impressions DESC
LIMIT 200

1B: Search Campaign Search Terms

SELECT campaign.name, search_term_view.search_term,
    metrics.impressions, metrics.clicks, metrics.cost_micros,
    metrics.conversions, metrics.conversions_value
FROM search_term_view
WHERE campaign.advertising_channel_type = 'SEARCH'
AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.impressions DESC
LIMIT 200

1C: Campaign Performance per Channel Type

SELECT campaign.name, campaign.advertising_channel_type,
    campaign.status, metrics.impressions, metrics.clicks,
    metrics.cost_micros, metrics.conversions,
    metrics.conversions_value, metrics.search_impression_share
FROM campaign
WHERE segments.date DURING LAST_30_DAYS
AND campaign.status = 'ENABLED'
ORDER BY metrics.cost_micros DESC

1D: PMax Channel-Level Breakdown (where the budget goes)

Available from API v22+: segments.ad_network_type works in PMax campaigns, enabling per-channel cost/conversion breakdown.

SELECT campaign.name, segments.ad_network_type,
    metrics.impressions, metrics.clicks, metrics.cost_micros,
    metrics.conversions, metrics.conversions_value
FROM campaign
WHERE campaign.advertising_channel_type = 'PERFORMANCE_MAX'
AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.cost_micros DESC

Network type values: SEARCH, SEARCH_PARTNERS, CONTENT (Display/YouTube), MIXED.

Step 2: Brand Traffic Analysis

Brand vs Non-Brand Classification

BRAND TRAFFIC IDENTIFICATION
============================

Step 1: Define brand terms
+-- Exact brand name + variations
+-- Brand + product combinations
+-- Common typos of brand name
+-- URL-based searches (domain.com)

Step 2: Classify PMax search terms
+-- BRANDED:     Contains brand name or variation
+-- NON-BRANDED: No brand reference
+-- AMBIGUOUS:   Could be either (e.g., product name = brand name)

Step 3: Calculate ratios
+-- Brand Traffic %  = branded impressions / total PMax impressions
+-- Brand Spend %    = branded cost / total PMax cost
+-- Brand Conv %     = branded conversions / total PMax conversions
+-- Brand ROAS       = branded conv value / branded cost

ROAS Inflation Calculation

ROAS INFLATION FRAMEWORK
========================

Blended PMax ROAS = Total conversion value / Total cost
Non-Brand PMax ROAS = Non-brand conversion value / Non-brand cost
Brand-Only ROAS = Brand conversion value / Brand cost

EXAMPLE:
+-- Blended PMax ROAS:   8.5x  <-- This is what Google reports
+-- Brand-Only ROAS:     15.0x <-- Brand traffic always converts well
+-- Non-Brand PMax ROAS: 3.2x  <-- THIS is your real PMax performance
+-- Search Campaign ROAS: 4.1x <-- Compare with this

INFLATION RATIO = Blended ROAS / Non-Brand ROAS
+-- <1.5x -> Minimal inflation
+-- 1.5-2.5x -> Significant inflation
+-- >2.5x -> Severe inflation, PMax claiming mainly brand conversions

Step 3: Decision Tree -- Brand Exclusions

BRAND EXCLUSION DECISION TREE
==============================

Do you have a dedicated Brand Search campaign?
|
+---> NO
|   +---> FIRST: Create Brand Search campaign
|       +-- Exact match brand keywords
|       +-- High budget (brand converts cheaply)
|       +-- Wait 2-4 weeks for data -> go to YES
|
+---> YES -> Analyze PMax brand traffic %
    |
    +---> <20% brand traffic in PMax
    |   +---> NO ACTION NEEDED
    |       +-- PMax is finding mostly new traffic
    |       +-- Monitor monthly
    |
    +---> 20-50% brand traffic in PMax
    |   +---> CONSIDER BRAND EXCLUSION
    |       +-- Compare: Brand Search CPA vs PMax brand CPA
    |       +-- If Search cheaper -> enable exclusion
    |       +-- If PMax cheaper -> monitor, no action
    |
    +---> 50-70% brand traffic in PMax
    |   +---> BRAND EXCLUSION RECOMMENDED
    |       +-- PMax ROAS is misleading
    |       +-- Brand Search campaign is more efficient
    |       +-- Implement exclusion (see action recipes)
    |
    +---> >70% brand traffic in PMax
        +---> BRAND EXCLUSION REQUIRED
            +-- PMax is functioning as expensive Brand campaign
            +-- Non-brand ROAS is probably low
            +-- Also consider: is PMax even valuable?

Step 4: Setting Up Brand Exclusions

Via google_ads_mutate -- Campaign-Level Brand Exclusions

Add negative keywords to PMax campaign (Recipe #7):

operations=[
    {"campaignCriterionOperation": {"create": {
        "campaign": "customers/{CID}/campaigns/{PMAX_CAMPAIGN_ID}",
        "negative": true,
        "keyword": {"text": "[brand name]", "matchType": "PHRASE"}
    }}},
    {"campaignCriterionOperation": {"create": {
        "campaign": "customers/{CID}/campaigns/{PMAX_CAMPAIGN_ID}",
        "negative": true,
        "keyword": {"text": "[brand variation]", "matchType": "PHRASE"}
    }}},
    {"campaignCriterionOperation": {"create": {
        "campaign": "customers/{CID}/campaigns/{PMAX_CAMPAIGN_ID}",
        "negative": true,
        "keyword": {"text": "[brand typo]", "matchType": "BROAD"}
    }}}
]

Via Shared Negative Keyword List (Recipe #9)

Better for multiple PMax campaigns:

operations=[
    {"sharedSetOperation": {"create": {
        "resourceName": "customers/{CID}/sharedSets/-10",
        "name": "Brand Exclusions - PMax", "type": "NEGATIVE_KEYWORDS"
    }}},
    {"sharedCriterionOperation": {"create": {
        "sharedSet": "customers/{CID}/sharedSets/-10",
        "keyword": {"text": "brand name", "matchType": "PHRASE"}
    }}},
    {"sharedCriterionOperation": {"create": {
        "sharedSet": "customers/{CID}/sharedSets/-10",
        "keyword": {"text": "brand.com", "matchType": "BROAD"}
    }}},
    {"campaignSharedSetOperation": {"create": {
        "campaign": "customers/{CID}/campaigns/{PMAX_CAMPAIGN_ID}",
        "sharedSet": "customers/{CID}/sharedSets/-10"
    }}}
]

Step 5: Incrementality Testing Protocol

2-Week On/Off Test

INCREMENTALITY TEST PROTOCOL
==============================

GOAL: Determine if PMax actually delivers incremental conversions
      or only claims existing branded traffic.

WEEK 1-2: BASELINE (PMax ON)
+-- Record: Total account conversions
+-- Record: Search campaign conversions
+-- Record: PMax conversions
+-- Record: Total revenue

WEEK 3-4: TEST (PMax OFF)
+-- Pause PMax campaign
+-- Do NOT increase Search budget (important!)
+-- Measure: Total account conversions
+-- Measure: Search campaign conversions
+-- Measure: Total revenue

ANALYSIS:
+-- Total conversions dropped >20%?
|   +---> PMax delivers incremental value -> turn back on
+-- Total conversions dropped 5-20%?
|   +---> Mixed result -> PMax with brand exclusions
+-- Total conversions same or increased?
|   +---> PMax was only claiming brand traffic
+-- Search conversions increased while total stayed same?
    +---> Confirmed: PMax was cannibalizing Search

Statistical Significance Checklist

SIGNIFICANCE REQUIREMENTS
========================

[ ] Minimum 100 conversions per phase (baseline + test)
[ ] Minimum 14 days per phase (seasonal effects)
[ ] No major external factors (sales, holidays, PR)
[ ] No other campaign changes during test
[ ] Budget kept constant
[ ] No new competitor activity

RELIABILITY:
+-- <100 conversions/phase -> Not reliable, extend test
+-- 100-500 conversions/phase -> Indicative
+-- >500 conversions/phase -> Statistically reliable

Step 6: Monitoring After Exclusions

GAQL Queries for Ongoing Monitoring

After setting brand exclusions, monitor weekly:

-- PMax performance after exclusions (compare with baseline)
SELECT campaign.name, metrics.impressions, metrics.clicks,
    metrics.cost_micros, metrics.conversions,
    metrics.conversions_value
FROM campaign
WHERE campaign.advertising_channel_type = 'PERFORMANCE_MAX'
AND segments.date DURING LAST_7_DAYS

-- Search campaign performance (should increase)
SELECT campaign.name, metrics.impressions, metrics.clicks,
    metrics.cost_micros, metrics.conversions, metrics.conversions_value
FROM campaign
WHERE campaign.advertising_channel_type = 'SEARCH'
AND segments.date DURING LAST_7_DAYS

Expected Results After Brand Exclusion

WHAT TO EXPECT AFTER BRAND EXCLUSION
======================================

PMax Campaign:
+-- Impressions: -30 to -60% (normal, branded volume gone)
+-- Clicks: -20 to -50%
+-- ROAS: DROPS (this is the real non-brand ROAS)
+-- Conversions: -20 to -40%
+-- CPA: INCREASES (non-brand is more expensive, this is normal)

Search Campaigns:
+-- Impressions: +10 to +30% (brand traffic recaptured)
+-- Clicks: +15 to +40%
+-- ROAS: Same or INCREASES
+-- Conversions: +10 to +25%
+-- CPA: Same or DECREASES

Total Account:
+-- Conversions: Same or +5-10% (more efficient budget use)
+-- ROAS: Same or better (fairer distribution)
+-- CPA: Same or better

2026 Updates: PMax Negative Keywords + Search Term Reporting

• Campaign-level negative keywords are now supported in PMax (API v20+). Use google_ads_mutate Recipe #7 with campaignCriterionOperation — no longer limited to account-level negative lists.
• Search term reporting for PMax is available (API v21+): query search_term_view with PERFORMANCE_MAX filter (Step 1A above). This was not possible before v21 — if you had old reports showing no data, rerun.
• url_expansion_opt_out was removed in v22; URL expansion is now controlled via asset group settings.
• Asset performance labels (Best/Good/Low) were removed from the API in v22 — do not reference them in scripts.

Benchmarks & Red Flags

Metric	Healthy	Warning	Critical
Brand traffic % in PMax	<20%	20-50%	>50%
PMax ROAS / Non-Brand ROAS ratio	<1.5x	1.5-2.5x	>2.5x
Search impressions after PMax launch	Stable	-10 to -25%	>-25%
PMax search term relevance	>70% relevant	50-70%	<50%
Brand CPA: PMax vs Search	Comparable	PMax 1.5x more expensive	PMax 2x+ more expensive

Common Patterns

PATTERN 1: "Fantastic PMax ROAS"
+-- Symptom: PMax ROAS 10x+, Search ROAS 3-4x
+-- Cause: PMax claims branded last-click conversions
+-- Diagnosis: Check search terms -> >60% branded
+-- Solution: Brand exclusion -> real ROAS becomes visible

PATTERN 2: "Search performance declining"
+-- Symptom: Search impressions/clicks drop after PMax launch
+-- Cause: PMax wins branded auctions (higher Ad Rank)
+-- Diagnosis: Compare Search metrics before/after PMax
+-- Solution: Brand exclusion on PMax or priority structure

PATTERN 3: "PMax scales budget but no growth"
+-- Symptom: Higher PMax spend, but account ROAS drops
+-- Cause: PMax scales in branded traffic, not new customers
+-- Diagnosis: Calculate non-brand ROAS + incrementality test
+-- Solution: Brand exclusion + budget cap on PMax

PATTERN 4: "Conversion Value Rules not set"
+-- Symptom: All conversions weigh equally
+-- Cause: Brand conversions have different value than non-brand
+-- Diagnosis: Check if CVR differs per traffic source
+-- Solution: Set up value rules (Search Console scope)

Tools Reference

Action	Tool	Recipe/Query
Get PMax search terms	google_ads_run_gaql	search_term_view + PERFORMANCE_MAX filter
Search campaign terms	google_ads_run_gaql	search_term_view + SEARCH filter
Campaign performance per type	google_ads_run_gaql	campaign + advertising_channel_type
Set brand exclusion	google_ads_mutate	Recipe #7 (negative keywords)
Shared brand exclusion list	google_ads_mutate	Recipe #9 (shared negative keyword list)
Pause PMax (incrementality test)	google_ads_mutate	Recipe #2 (pause campaign)
Look up accounts	google_ads_list_accounts	--