Listen to this article · 8 min listen

Identifying underperforming elements within a Performance Max campaign isn’t just about tweaking bids; it’s about forensic analysis. Many marketers simply let these campaigns run, hoping the algorithm sorts itself out. That’s a mistake. Without deep dives into asset group performance and audience signals, you’re leaving money on the table, and worse, you’re often spending it inefficiently. How do you pinpoint the precise components dragging down your campaign?

Key Takeaways

  • Regularly audit Performance Max asset groups, specifically focusing on assets with “Low” or “Poor” ratings, as these are direct indicators of underperformance.
  • Analyze Cost Per Conversion (CPC) and Return on Ad Spend (ROAS) at the asset group level, not just campaign-wide, to isolate inefficient spending.
  • Segment audience signals by conversion rate to identify which audience types are truly driving results versus those consuming budget without converting.
  • Prioritize replacing or improving creative assets (images, videos, headlines, descriptions) that show low Click-Through Rates (CTR) or high bounce rates, as these directly impact engagement.
  • Implement A/B testing for new asset variations and track their performance against existing underperformers before making widespread changes.

I recently managed a Performance Max campaign for an e-commerce client selling specialized outdoor gear. The campaign aimed to drive online sales for a new line of hiking boots. Our budget for this initiative was $15,000 per month, running for a three-month duration. The initial setup involved broad targeting, relying heavily on Google’s automation to find converting customers.

The campaign’s overall metrics after the first month looked acceptable on the surface: a Cost Per Lead (CPL) of $35 and a ROAS of 2.1x. However, the client expected a ROAS closer to 3.0x, aligning with their profit margins. This discrepancy signaled a problem, despite the seemingly “okay” averages. Digging deeper was essential.

Our strategy was straightforward: showcase high-quality product imagery and compelling calls to action across all available Google channels. We used a mix of audience signals, including custom segments based on competitor websites and in-market audiences for outdoor recreation. The creative approach focused on aspirational lifestyle shots of hikers using the boots in rugged terrain, coupled with benefit-driven headlines highlighting durability and comfort. We supplied a full complement of assets: 20 images, 5 videos, 15 headlines, and 5 descriptions per asset group, aiming for maximum algorithmic flexibility.

What worked initially? Certain product images with individuals actively hiking performed well, achieving a Click-Through Rate (CTR) of 1.8% on display networks. Our short, punchy headlines also saw higher engagement, particularly those emphasizing “all-weather protection.” The campaign generated 1.5 million impressions in its first month, leading to 428 conversions. The average cost per conversion was around $35.05.

However, the campaign’s overall ROAS lagged. This wasn’t a problem with the product or the market; it was an execution issue. The first step in diagnosing this was to examine the “Asset Group” performance report within the Google Ads interface. This report is critical, offering a granular view often overlooked by those who only glance at campaign-level data. We noticed a significant disparity in asset group performance. One asset group, targeting “adventure travelers,” had a ROAS of 1.2x, while another, focused on “serious hikers,” achieved 2.8x. This alone pointed to inefficient spending.

Within the underperforming “adventure travelers” asset group, I then accessed the “Assets” report. This is where the real insights emerged. Google provides performance ratings for individual assets: “Best,” “Good,” “Low,” or “Poor.” We found several videos and images rated “Low” or “Poor.” Specifically, a video showing only static product shots, without any human interaction, had a CTR of 0.3% and was contributing to a disproportionately high number of non-converting impressions. This video alone had accumulated over 200,000 impressions but zero conversions attributed directly to its views. That’s a clear indicator of a drain on the budget.

Similarly, several headlines were rated “Low.” These were often generic phrases like “Shop Our Boots” which lacked any compelling value proposition. These headlines had an average CTR of 0.5% when paired with various descriptions, suggesting they failed to capture user interest. We also observed a high bounce rate (over 70%) for traffic originating from specific ad combinations featuring these “Low” rated assets, according to our Google Analytics data.

The audience signals also needed scrutiny. While the “serious hikers” audience performed well, the “adventure travelers” segment, which included a broader demographic, was less efficient. When we segmented conversions by audience type, we saw that while “adventure travelers” generated a respectable volume of clicks, their conversion rate was nearly half that of “serious hikers.” This indicated that while the system was finding clicks, it wasn’t finding the right kind of clicks for this particular segment.

My optimization steps were methodical:

  1. Asset Replacement: Immediately paused all “Low” and “Poor” rated videos and images. Replaced the static product video with a new one featuring dynamic action shots of hikers.
  2. Headline Refinement: Rewrote the “Low” performing headlines to be more benefit-oriented. Instead of “Shop Our Boots,” we tested “Tackle Any Trail: Unrivaled Grip & Comfort” and “Lightweight Durability for Your Next Summit.”
  3. Audience Signal Adjustment: Narrowed the “adventure travelers” audience signal. Instead of broad interests, we focused on custom segments derived from search terms related to specific hiking boot brands and technical outdoor apparel. This was a critical adjustment; broad signals can easily lead the algorithm astray.
  4. Budget Reallocation: Shifted a small portion of the budget (around 10%) from the underperforming “adventure travelers” asset group to the higher-performing “serious hikers” group, testing the impact.

After implementing these changes over the next two weeks, we saw a noticeable improvement. The campaign’s overall ROAS climbed to 2.6x. The “adventure travelers” asset group, post-optimization, saw its ROAS increase to 1.9x. The new video asset achieved a CTR of 1.2%, a significant improvement over the 0.3% of its predecessor. The refined headlines also pushed CTRs above 1.0% on average for that asset group.

The key here was not just identifying the symptoms (low ROAS), but diagnosing the root cause at the most granular level possible: individual assets and specific audience signals. Performance Max, while powerful, requires vigilant oversight. You cannot simply set it and forget it, especially when initial results fall short of expectations. The system needs constant feedback, and that feedback comes from identifying and replacing its weakest links. Ignoring these underperformers is like trying to win a race with flat tires on your car. You’ll move, but never efficiently.

Many marketers assume Performance Max is a black box. It isn’t entirely. The asset reports and audience signal insights provide a window into its decision-making. My advice is to trust the data, not just the “automation.” If an asset is rated “Low,” it’s costing you. If an audience signal yields low conversion rates, it’s wasting budget. Act on that information. It’s a continuous process of refinement, not a one-time setup.

Regularly auditing your Performance Max asset groups and adjusting based on specific data points will always yield better results than simply hoping for the best. Focus on replacing or improving the weakest assets and refining your audience signals to ensure the campaign reaches the right people with the most compelling message.

How frequently should I review Performance Max asset group performance?

You should review asset group performance at least weekly, especially for campaigns with significant budgets or those in their initial learning phases. For established campaigns, a bi-weekly review can suffice, but always react promptly to any sudden dips in ROAS or increases in Cost Per Conversion.

What is a “Low” or “Poor” asset rating, and what should I do about it?

“Low” or “Poor” asset ratings from Google Ads indicate that the specific creative (image, video, headline, description) is underperforming compared to other assets in your campaign. You should pause these assets immediately and replace them with new, higher-quality variations. Analyze why they performed poorly; often, it relates to relevance, clarity, or visual appeal.

Can I see which specific channels (Search, Display, YouTube) my Performance Max ads are running on?

While Performance Max aggregates reporting, you can gain insights into channel distribution through the “Placement” report under “Where ads showed” in Google Ads. This report details where your ads appeared, including specific websites, apps, and YouTube channels, giving you a directional sense of channel performance, though not individual asset performance per channel.

Is it possible to exclude certain placements in Performance Max if they are underperforming?

Yes, you can exclude specific placements (websites, apps, YouTube channels) in Performance Max. If the “Placement” report shows certain placements are consuming budget without conversions, you can add them to a “Negative Placements” list at the account level. This is a crucial step for preventing wasteful spending on irrelevant or low-quality traffic sources.

How do audience signals impact Performance Max campaign performance?

Audience signals guide the Performance Max algorithm towards users most likely to convert. If your signals are too broad or irrelevant, the campaign may spend budget on less qualified audiences, leading to lower conversion rates and ROAS. Refining these signals to include highly relevant custom segments, first-party data, or precise in-market audiences is essential for directing the algorithm effectively.