● PMAX PLACEMENT EXCLUSIONS

Cut the junk.
Keep the engine.

A do-this-now brief for the performance team: stop PMax wasting spend on mobile-game & arbitrage placements — without ever touching the Shopping inventory that drives the account.

SCOPE Account 607-429-8123NETWORK Performance MaxRISK <1 conv/moREVERSIBLE Yes
0
of impressions = Shopping (the engine — untouched)
0
Display cost / conversion (vs ₹320 Shopping)
0
max conversions/month at risk
0
junk placements to exclude (137 apps + 2 sites)
Download exclusion list (CSV)
01 THE VERDICT

Three things, then act.

✓What

Add the attached list as an account-level placement exclusion. PMax stops serving on junk apps & arbitrage sites.

⚡Why it's safe

Shopping (80.7% of delivery) can't be touched by placement exclusions. Worst case is <1 conversion/month, fully reversible.

◎How to check

Baseline the Display (CONTENT) network, apply, re-check in 7 days. Win = cost down, conversions flat.

02 THE PROBLEM

Where PMax actually served

30-day placement report. Four-fifths is the Shopping engine. The waste is the ~19% running on the Display network — in-app and random sites.

80.7%
14.7%
Shopping / Google-owned — keep Display websites In-app Display (4.6%)

What that Display spend buys — pulled live from the account:

Tinder8 Ball PoolWord CookiesSnake ClashPhone CleanerGPS Map CameraAaj Tak / Way2NewsIRCTCzexoads.comfastdl.app

Games, utilities, news, dating, train apps + arbitrage sites. Zero art-buying intent — classic accidental-tap inventory.

03 WHY IT'S SAFE

The numbers, not a hunch

Display is the account's weakest network — and too small to be hiding real sales.

Network (Mar–Jun)SpendCost / conv
Shopping / Search
= 80.7% · the engine₹678k₹320
Display (CONTENT)
= where junk lives₹12.7k₹622

Max-downside calculator

Whole non-Shopping Display footprint = 46,279 impressions/mo. Drag the worst-case assumptions — the answer stays tiny.

≤2.3
conversions per month at risk — across every excluded placement
2.3 of ~525 monthly Search/Shopping conversions · 0.4%

Mobile apps alone (the highest-confidence cut) cap at <1 conversion/month. And Shopping is untouched regardless — placement exclusions only filter Display/video.

Honest caveat for the team: Google doesn't expose cost/conversions per placement for PMax — only impressions. So this relies on network-level data + a bounded worst-case + reversibility, not per-placement ROAS (which doesn't exist for PMax).
04 WHAT TO DO

Exclude these. Leave those.

Tier A Exclude now — near-zero risk ▾
All 137 mobile apps (games, utilities, news, dating, transport) + the two arbitrage domains. This is the attached CSV.
137 mobile appszexoads.comfastdl.app
Tier C Do NOT exclude — the trap ▾
Trivial impressions, but a real (if rare) buyer could hide here. Leaving them costs almost nothing.
bbc.comtheprint.indeccanherald.comscribd.comlearncbse.inecosia.orgolx.in

Do not blanket-exclude "all websites."

Never Shopping / GOOGLE_PRODUCTS ▾
80.7% of delivery and the engine. Placement exclusions can't reach it anyway — just never attempt to.
05 HOW TO IMPLEMENT — THE DURABLE WAY

Account-level. Always.

Critical: apply at account level. Campaign-level placement exclusions are silently ignored by Performance Max — they will do nothing.
Add the list at account level
UI: Admin/Tools → Account-level placement exclusions → paste the paste_value_UI column. Or Google Ads Editor → account-level negative placements. (API: CustomerNegativeCriterion.)
Add a mobile-app category exclusion
Listing individual apps is whack-a-mole. An account-level app-category exclusion auto-catches future junk apps — no monthly re-upload.
Automate the review (optional, best)
A monthly Google Ads Script that pulls performance_max_placement_view and auto-excludes new app/arbitrage placements. Removes the manual chore entirely.
06 HOW TO DOUBLE-CHECK

Cost down, conversions flat

Run this before applying and again 7 days after. Watch the CONTENT row.

GAQL · network split (run before + after)
SELECT segments.ad_network_type, metrics.cost_micros,
       metrics.conversions, metrics.conversions_value
FROM campaign
WHERE campaign.advertising_channel_type = 'PERFORMANCE_MAX'
  AND segments.date DURING LAST_30_DAYS
✓Pass

CONTENT cost is down, conversions flat (any drop ≤ ~1/week — that's the bound).

!Investigate

CONTENT conversions fall > ~1/week, or total PMax conv/day dips → roll back (instant, no learning reset).

GAQL · confirm junk is gone
SELECT performance_max_placement_view.placement_type,
       performance_max_placement_view.display_name, metrics.impressions
FROM performance_max_placement_view
WHERE segments.date DURING LAST_14_DAYS
ORDER BY metrics.impressions DESC
07 ROLLOUT CHECKLIST

Tick as you go

Saved on this device — pick up where you left off.

0%
Baseline the CONTENT network Run the GAQL split, note cost + conversions
Upload Tier A at account level The CSV — apps + zexoads.com + fastdl.app
Add mobile-app category exclusion The durable, anti-whack-a-mole layer
Wait 7 days, re-run the split Confirm: cost down, conversions flat
Confirm junk gone + schedule monthly review Or ship the auto-exclusion script
08 QUICK ANSWERS

Before you ask

Won't this hurt Shopping / PMax performance?
No. Placement exclusions only filter Display/video inventory. Shopping & Search (GOOGLE_PRODUCTS, 80.7% of delivery) are untouched by construction.
Will it reset PMax learning?
No — unlike pausing a campaign, adding/removing placement exclusions does not reset the learning phase. It's reversible in minutes.
Why not just exclude all Display?
Display still produced ~20 conversions in 4 months, and a few contextual placements convert. Surgical > nuclear — exclude the named junk, keep Tier C.
Google "conversions" look inflated.
They are — they include view-through, store visits and some double-counting (~2–3× vs actual orders). Treat all ROAS/conv figures here as relative, not absolute. MAPS conversions = store visits, not online sales.
Copied to clipboard ✓