How to Analyze PMax Asset Performance in Google Ads

PMax asset performance is best analyzed as a layered system, not a leaderboard of headlines, images, and videos.

Start with the campaign goal, then move through channel performance, asset groups, individual assets, and combinations. Each layer answers a different question. If you skip directly to a row with a high CPA or low conversion count, you can remove an asset that helped a useful combination serve.

The core limitation is attribution: Performance Max assembles ads from multiple assets across Google inventory. One conversion is not cleanly divisible into “this headline caused 40% and this image caused 60%.” Google’s current reporting gives much more detail than it used to, but the data still needs careful interpretation.

The five layers of Performance Max asset analysis

Layer	Best question	Do not use it to claim
Campaign	Is PMax achieving the business goal at an acceptable marginal return?	Every asset group or channel must match the campaign average.
Channel	Where is the campaign serving and contributing results?	A channel with less volume is inherently bad.
Asset group	Which audience, product, service, landing page, or creative theme is represented?	One asset group caused the campaign’s entire change.
Individual asset	Which text, image, or video deserves investigation or a new challenger?	The asset performed independently of every combination.
Combinations	Which assembled messages appear among the top combinations?	The report is an exhaustive A/B test of every possible combination.

This hierarchy prevents a common PMax mistake: using a narrow diagnostic signal to make a broad budget or campaign decision.

Step 1: validate the campaign before the creative

Asset analysis is downstream of measurement and economics. Before reviewing creative, confirm:

• Primary conversion actions reflect valuable outcomes.
• Conversion values are credible if the campaign uses value-based bidding.
• Conversion lag is included in the date range.
• Lead spam, duplicate actions, or imported offline outcomes are handled consistently.
• Budget, targets, feed, landing pages, and major settings did not change unexpectedly.
• The selected period contains normal demand rather than a one-off promotion or outage.

If conversion tracking is wrong, asset metrics will be precise descriptions of the wrong outcome.

Step 2: inspect campaign and channel performance

At campaign level, review cost, conversions, conversion value, CPA or ROAS, and the actual business outcome. Use the campaign as the primary unit because Performance Max bidding optimizes toward campaign goals.

Then open Insights and reports → Channel performance. Google’s Performance Max channel performance report shows how the campaign is serving across available channels and includes campaign-level summary metrics, channel distribution, diagnostics, and time-series views.

Use this layer to add context:

• A text asset serving mostly with Search inventory has a different job from one supporting a video or Display combination.
• A sudden shift in clicks or conversions may reflect a channel mix change, not a copy change.
• No delivery on one channel can mean the campaign is currently finding better return elsewhere; it is not automatically an error.

Channel data explains the environment in which the assets participated. It does not convert PMax into separate manually optimized channel campaigns.

Step 3: compare asset groups by business theme

Go to Campaigns → Asset groups, choose the PMax campaign, switch from Summary to Table, and add the metrics relevant to the goal.

The Performance Max asset group report can show conversions, conversion value, and other performance columns. Google recommends continuing to judge overall success at campaign level because an asset group with a higher average CPA or lower average ROAS can still contribute incremental results to the campaign goal.

An asset group comparison is most useful when each group has a coherent meaning, such as:

• One product category
• One service line
• One landing-page family
• One audience need
• One promotional theme

If several groups contain the same products, URLs, and generic creative, a performance difference may be difficult to translate into an action. Improve the structure before expecting the report to explain the business.

Questions to ask at asset-group level

1. Does the asset group represent one clear offer or intent?
2. Do the landing page and product data match that theme?
3. Is conversion quality materially different from other groups?
4. Did the group have enough time and eligible inventory to serve?
5. Is a policy, feed, or asset-status issue limiting delivery?
6. Would removing the group eliminate useful incremental reach?

The goal is not to force every asset group to the same average. It is to understand the role and identify a correctable input.

Step 4: analyze individual PMax assets

In Google Ads, select Assets from the Campaigns menu and choose Assets from the view dropdown. Filter to the relevant Performance Max campaign or asset group, then add columns for the metrics you need.

Google’s asset-level metrics documentation lists impressions, clicks, cost, conversions, and conversion value for Performance Max assets. That additional transparency is useful, with two important caveats.

Asset metrics do not sum to campaign totals

If an ad impression contains a headline, image, and video, each served asset can receive an impression. The same principle applies to clicks and other metrics. Adding the rows creates double- or triple-counting rather than a campaign total.

A conversion can be credited to multiple assets

Google’s conversion reporting by asset guide explains that a conversion is not split among the components of the ad. If a converting ad included a headline, description, and image, all three can receive credit for the conversion.

Therefore:

• Use impressions to understand exposure.
• Use clicks, conversions, and value to identify patterns worth investigating.
• Treat CTR, CPC, CPA, and ROAS as directional within a fair scope.
• Do not call the top conversion row the causal winner.

Step 5: review top asset combinations

Open the PMax campaign, select Asset groups, choose View asset details, and open the Combinations tab.

Google’s combinations report shows the top six combinations in each of three categories: text, image, and video. It can also preview how combinations appear across channels. It is valuable for questions such as:

• Which value propositions repeatedly appear together?
• Does the resulting ad communicate one coherent offer?
• Are repeated words making the combination awkward?
• Do image and text assets reinforce the same product or use case?
• Is critical context absent from the top combinations?

It does not show every possible combination or prove the incremental value of one component. Use it to understand assembled creative, not to reverse-engineer the entire delivery model.

A practical PMax asset decision model

Evaluate each candidate change against four dimensions.

1. Eligibility and accuracy

Replace or remove an asset promptly if it is disapproved, factually wrong, expired, off-brand, or sends a promise the landing page cannot support.

This decision does not require a statistical performance case. Accuracy is a release requirement.

2. Message contribution

Label the job of the asset: relevance, benefit, proof, offer, objection handling, urgency, or call to action.

An asset with modest volume may still provide a unique message that allows the system to answer a narrower intent. Before removing it, check whether another eligible asset can perform the same job.

3. Directional performance

Within the same asset group, asset type, date range, and broadly similar serving context, look for repeated evidence instead of one volatile ratio.

An investigation queue might include:

• High-exposure text with weak engagement and a vague proposition
• An asset with conversions but materially worse efficiency across a stable period
• An asset that rarely serves and duplicates a stronger message
• A top combination that produces an unclear or contradictory assembled ad

4. Replacement quality

Do not remove a weak asset until you have a better, eligible alternative. A replacement should add a distinct and accurate message, not another paraphrase of the current winner.

Google’s current Performance Max text asset guidance recommends waiting roughly two to three weeks after upload before checking Ad Strength and asset group status. The right performance window can be longer when conversion volume is low or lag is substantial.

For the writing side of the process, see how to optimize Performance Max headlines and descriptions.

Example: turn the reports into one decision

Suppose a local HVAC PMax campaign contains an asset group for emergency repairs.

You find:

• Campaign CPA is within target, but qualified-lead rate declined.
• Channel performance shows more Search interactions during the decline.
• The emergency-repair asset group received most of that increase.
• A generic “Affordable HVAC Services” headline has high exposure and weak lead quality.
• Top combinations pair it with a broad description and a same-day-service image.

The evidence does not prove the generic headline caused low-quality leads. It supports a narrower hypothesis: the assembled message may be attracting price-first traffic without clarifying emergency intent.

A useful challenger would add qualification, such as “Emergency AC Repair Today,” while keeping the landing page, bidding, and other major inputs stable. The evaluation should use qualified leads and cost per qualified lead, with total conversions as a guardrail.

That is a defensible asset decision. “Delete every headline with a high asset-level CPA” is not.

When to use an experiment

Use routine iteration for accuracy fixes, obvious duplication, and low-risk message improvements. Use an experiment when the question is strategically important and the campaign has enough volume.

Google Ads now supports Performance Max A/B asset testing in beta for eligible campaigns, allowing different asset sets within one asset group to receive separate traffic splits. Google recommends a four-to-six-week minimum for that test type and locks the tested asset group from asset edits while it runs.

Availability and eligibility can vary. If the option is unavailable, document the change window and interpret the before-and-after result as directional rather than randomized evidence.

Use one review queue across matching text

The same headline or description can appear in more than one place. Exporting raw rows can make the text look like several separate assets even when the business question is “How is this message performing within this campaign?”

LeadUp’s PMax asset analysis workflow lets you navigate campaign and asset-group scope, combine performance for matching text within the selected scope, and review an AI-assisted rewrite beside the current asset. It keeps the final decision with the account manager.

Final takeaway

Good Performance Max asset analysis moves from broad outcome to narrow hypothesis:

1. Validate conversions and campaign economics.
2. Use channel data to understand the serving context.
3. Compare asset groups by coherent business theme.
4. Read individual metrics directionally and never sum asset rows.
5. Inspect top combinations before rewriting a component.
6. Replace one message with a genuinely better alternative.
7. Use an eligible experiment when the decision is important enough to prove.

The reports are most valuable when they help you make a smaller, clearer decision, not when they tempt you to attribute an entire campaign result to one headline.