Skills

pmax-asset-group-reviewer

Google Shopping v2
@adsledger 97 installs updated 7d ago MIT license

PMax asset group reviewer

Performance Max hides most of its levers, so structure is what you control. This skill checks that each asset group has a clear product set, a matching message and enough assets to run.

When to use

  • After a new PMax launch has run 3–4 weeks.
  • When PMax spend grows but Shopping-like ROAS falls.

Input

Per campaign: budget, bidding (tROAS / max conversion value), target, and brand exclusions. Per asset group: listing group tree, audience signals, final URL expansion setting, asset counts (headlines, long headlines, descriptions, images, logos, videos) and asset-level performance ratings. Product-level performance by asset group for 30 days.

Checks

  1. Listing groups. Each asset group should subdivide by one dimension (custom label, product type, brand) and cover a coherent set. Flag an "All products" group sitting next to specific ones in the same campaign; flag products appearing in two campaigns, since only one will serve.
  2. Assets. Minimum to run well: 5 headlines (up to 15), 1–5 long headlines, 2–5 descriptions (one ≤ 60 chars), landscape and square images, a logo, and a real video (otherwise Google auto-generates one). Flag "Low" rated assets older than 30 days.
  3. Message match. Headlines should name the product set in the listing group; a generic brand message on a category group is a finding.
  4. Signals. Customer-list and search-theme signals should differ per group; identical signals everywhere add nothing.
  5. Final URL expansion and brand. Note whether expansion is on and whether brand terms are excluded; report, do not judge, if the store has no stated policy.
  6. Zombies. Share of products with zero impressions per group; above 40% suggests a separate zombie campaign.

Rules

  • Read-only. Recommendations only; no asset, listing group or budget change is made.
  • Asset-level conversion data is not available in PMax; do not invent it. Use ratings and product-level data.
  • Missing exports are listed at the top of the report.

Output format

Campaign "PMax – Footwear" · tROAS 400% · 30d ROAS 3.1
| asset group | products | zero-impr. | headlines | video | finding |
| Trail – m-high | 184 | 22% | 11 | yes | ok |
| All products | 2 040 | 61% | 5 | auto | catch-all overlaps Trail; split by label_1 |
Next: move 1 240 zombie products to a separate low-budget campaign.

License

MIT