PMax assets, scored in seconds.
PMax Asset Diagnostics runs a rule-based audit on any Performance Max campaign in your account. Scores the asset library 0–100, flags missing asset types, names the asset groups dragging the campaign down, and writes a suggested fix per finding. Read-only — nothing changes in your account.
One PMax campaign, all its asset groups.
Each run is scoped to one PMax campaign. The tool reads asset-group composition, asset-type counts, ad-strength labels, and Google-supplied performance labels — no writes — and returns a structured report. These are the numbers behind one sample run.
Out: a PMax health readout , per asset group.
Below is a sample readout from the run above. Overall health score, severity-banded findings count, and the top individual findings — each with the affected asset group, the rule that fired, and a suggested fix.
Each check is a rule , not a model.
Diagnostics is rule-based, not ML. Each finding has a predicate against PMAX_ASSET_LIMITS or a Google-supplied label. Same inputs, same findings, same severity — every run.
Counts vs PMAX_ASSET_LIMITS.
For each asset type — headlines, long headlines, descriptions, marketing images, square images, logos — we count what's there and compare to the required minimum. Missing a required type fires critical; thin headline/description counts fire warnings. A type already flagged as missing-critical is not also counted as a low-count warning.
Google's label distribution, not ours.
We read Google's POOR / AVERAGE / GOOD / EXCELLENT label per asset, as of the latest day in the window. We don't compute our own creative score. A POOR present is critical; an AVERAGE-only group (no GOOD or EXCELLENT) is a warning.
Google's LOW label, as of latest day.
We read Google's asset-level performance label (BEST / GOOD / LOW / LEARNING) as of the most recent day — not an aggregate. Assets currently labelled LOW are flagged for replacement.
Eligibility from Google's primary status.
We read Google's primary serving status per asset. Disapproved / not-eligible assets are critical (not serving at all); limited assets are a warning (serving with restricted reach). Benign approved-but-labelled assets are not flagged.
0–100 with hardcoded penalties.
Start at 100. Each critical finding deducts 25 points; each warning deducts 10. Healthy asset groups don't add — the score reflects what's broken, not what's working. Floor at 0.
Fixed fix strings, daily + on-demand.
Suggested actions are template strings tied to each check — not Lyra-Agent-written, by design, since the rules are crisp enough to name the fix without narration. A daily background pass runs the audit fleet-wide; on-demand single-campaign runs return in well under a second.
PMax Asset Diagnostics exposed asset-level coverage gaps and three theme mismatches. Qualified lead rate recovered from 18% to 26% the next quarter.
The honest answers about Diagnostics.
Is this AI?
Detection is rule-based — same inputs, same findings, same severity. Diagnostics doesn't use an LLM internally; suggested actions are template strings tied to each rule. Other Lyra tools use the Lyra Agent for narrative generation; this one doesn't need it because the fixes are crisp.
How often does the audit run?
Both ways. A daily background pass runs the audit across your account so findings land in your recommendations feed without you asking. You can also run it on-demand against any PMax campaign — point the tool at it and findings come back in well under a second, ideal right after an asset upload or before a quarterly review.
Does it modify my Google Ads account?
Never. Diagnostics has read-only scope. The output is a scored report with template-string suggested actions; applying anything is a separate step you do in Google Ads or in Google Ads Editor.
What does ad strength come from — your model or Google's?
Google's. We read Google's POOR / AVERAGE / GOOD / EXCELLENT label per asset group straight from the API and report the distribution. We don't compute our own creative score; that's Google's call.
Does this work on Search or Shopping campaigns too?
Not directly — this tool is scoped to Performance Max. For Search and Shopping campaign-level health, run Campaign Health Analyzer; it covers structure, spend, performance, Quality Score, and conversion across all campaign types.
Run the audit.
Keep the findings.
Connect your account and let Lyra run its 18 tools for 14 days. If the projected waste recovery isn't worth at least 10× the $49, don't pay. You keep every insight either way.
Read-only connect · write access opt-in per tool · SOC 2 in progress · GDPR + CCPA compliant