Fix: Add the PMax channel performance report to your weekly review

The Performance Max channel breakdown has been generally available since April 2025, but most weekly review cadences still skip it — partly because nothing breaks if you ignore it, partly because the report rewards looking at it the same way every week, not deeply once a quarter. This rule fires when your account does not have a documented review habit for the channel performance card; the fix is a five-minute weekly check, not a research project.

Why this matters

The channel performance card is the only ongoing surface that tells you whether PMax is quietly drifting toward inventory you would not have bought directly — YouTube placements that look like brand awareness on the bidder side but never convert, Search Partner traffic with lower CVR than google.com, Display spend on parked domains. None of that registers in your ROAS dashboard until two or three weeks after the fact, by which point Smart Bidding has trained on the drift. A weekly cadence catches the drift inside the same learning window it formed in.

The trap is treating the weekly review as a decision meeting. It is not — it is a watch-list update. The deeper interpretation work (and the explicit caveat that channel split is served-impression, not causal) belongs to [[pmax-channel-performance-timeline]]; this rule is purely about whether the review happens at all.

How to verify

In the Google Ads UI:

Campaigns → click into the PMax campaign → Insights tab → Channel performance card

Inside the card, you will see:

• A summary row (impressions, clicks, conversions, cost) split per channel
• A channel-to-goals flow chart (Sankey-style)
• A distribution table with channel-level metrics

For the weekly cadence, the distribution table is what you want. Screenshot it (or export to Sheets) with the same date window every week — a trailing 7-day or trailing 14-day window, your choice, but pick one and stick with it.

How to fix

1. Pick a fixed weekly cadence. Monday morning is the common choice — last week's data has fully attributed and Smart Bidding has settled. Block 5 minutes per PMax campaign.
2. Standardize the date window. Trailing 14 days is the safer default per the read-window guidance in [[pmax-channel-performance-timeline]]; trailing 7 days is acceptable for high-volume campaigns (>500 conv/week) where you actually need the cadence.
3. Log channel-conv-share and channel-spend-share for each of the six channels (Search, YouTube, Display, Discover, Gmail, Maps) in a sheet. One row per week per campaign. This is the watch-list.
4. Flag any of three red-flag patterns for action review:
   • Conv-share vs spend-share delta on any channel exceeds ~15 percentage points (steady state) or moves >15pp week over week
   • Search Partners share of Search-slice clicks exceeds 25%
   • YouTube spend share rises more than 10pp above its 4-week trailing average without a creative or asset change that explains it
5. Confirm at 14 days before acting. If a red flag persists on the next weekly check (i.e., not a single-week artifact), escalate into the diagnostic process documented in [[pmax-channel-performance-timeline]]. Do not change asset groups, exclusions, or bidding off a single weekly read.
6. Cross-reference Search slice with Brand campaign search terms monthly (not weekly — too slow-moving to be worth the weekly minute). This catches PMax cannibalizing branded queries; the fix path is brand exclusions, covered separately in [[fix-pmax-brand-exclusions]].

Edge cases

• New PMax campaigns under 14 days old: the channel mix is still stabilizing as Smart Bidding learns. Skip the review until day 14 and start the cadence from there.
• Very low volume campaigns (<50 conversions / week): the weekly read is noise. Move to bi-weekly or monthly cadence and use a trailing 28-day window.
• Campaigns paused mid-week: the partial week distorts channel-share math. Either exclude the week or note it explicitly in the log.
• Account uses portfolio bid strategies: channel mix can shift in response to other campaigns inside the portfolio, not anything you changed in this PMax. Note portfolio composition changes in the same log.

Industry benchmarks

The ~15pp delta threshold for the conv-share vs spend-share gap is the most-cited heuristic in agency walkthroughs of the report; it traces back to Mike Ryan's December 2025 Search Engine Land guide and is the same threshold used as the steady-state action trigger in [[pmax-channel-performance-timeline]] [2]. Search Partner share above 25% of Search-slice clicks is a softer heuristic but consistent across Adalysis and Optmyzr coverage of the post-April-2025 transparency rollout — when those clicks materially under-convert vs google.com, the 25% line is where the opt-out conversation starts.

Related

• [[pmax-channel-performance-timeline]] — the longer read on what the channel report actually shows and the served-impression vs causal-attribution distinction. Required reading before you act on any flag from the weekly review.
• [[fix-pmax-brand-exclusions]] — the fix path for PMax Search-slice cannibalization of branded queries.
• [[performance-max]] — the campaign type the report is scoped to.
• [[smart-bidding-learning-phase]] — why short-window reads of the channel timeline are noise during the first 14 days of any campaign or material change.

Sources

1. Google Ads Help — About the channel performance report for Performance Max. Confirms the card location (Insights → Channel performance), the six-channel split, and the available segments.
2. Search Engine Land — How to use Google's Channel Performance report for PMax campaigns, Mike Ryan, 2025-12-17. Source of the ~15pp conv-share vs spend-share delta heuristic and the caveat that the Sankey visualization can mislead at-a-glance reads.
3. Google blog — Channel performance reporting coming to Performance Max, 2025-04-30. Original rollout announcement; baseline framing for the report's intended use as a transparency surface.