Performance Max asset group strategy is the structural backbone that determines where Google spends your budget, what creative it serves, and which audiences it prioritizes. When asset groups are poorly structured, Performance Max campaigns silently redirect spend toward low-intent placements, particularly when auto-generated video assets fill gaps the advertiser never intended to leave open. This is the story of a mid-market ecommerce brand that watched 18 months of consistent ad spend produce steadily declining returns, traced the problem back to its PMax asset group architecture and auto-generated video problems, and rebuilt its way to meaningful revenue recovery without adding a dollar to total budget.
The brand sells consumer goods across roughly a dozen product categories, runs around $45K per month in Google Ads spend, and had been relying on Performance Max as its primary campaign type for nearly two years. What follows is what broke, why the obvious fixes kept failing, and what the rebuild actually looked like.
The Setup: A Mid-Market Ecommerce Brand With A Google Ads Ceiling
Account History: 18 Months Of Flat Performance On Consistent Spend
This brand had been advertising on Google for years before consolidating into Performance Max. The initial migration went well. ROAS held steady through the first few months, and the team felt comfortable letting the algorithm handle distribution across Search, Shopping, Display, YouTube, and Discovery placements.
But by month six, a pattern emerged: ROAS was stable, but revenue contribution from Google Ads was slowly declining. The account was spending the same amount and technically hitting target returns, but generating fewer total conversions at higher cost per acquisition. The dashboard looked fine at a glance. The trajectory underneath was not.
What The Account Looked Like: Campaign Structure, Bidding, Creative
The account ran three Performance Max campaigns, loosely organized by margin tier rather than product category or customer intent. Each campaign contained a single asset group with a broad mix of product listings, headlines, descriptions, and images pulled from across the catalog.
No custom video assets had been uploaded. Google's auto-generated video filled the gap, pulling from static product images, brand colors, and headline text to create generic slideshows that ran across YouTube and Display placements.
Bidding was set to maximize conversion value with a target ROAS, which is a reasonable default. But because the asset groups were so broad, the algorithm had limited signal about which products to push to which audiences.
The Symptoms: Stable ROAS, Declining Revenue Contribution, Rising CPAs
The clearest warning sign was the gap between ROAS and actual revenue growth. When ROAS stays high but revenue stays flat, the algorithm is usually optimizing for easy conversions rather than incremental ones. In this case, the brand was effectively paying more to reach people who would have converted anyway, while losing reach among new, high-intent buyers.
CPA crept up by roughly 15-20% over 12 months. Not dramatic enough to trigger alarms, but enough to erode margin on lower-AOV products. The team tried adjusting ROAS targets, pausing underperforming products, and refreshing ad copy. Nothing moved the needle in a lasting way.
The Diagnosis: Performance Max Overspending In The Wrong Places
How Asset Group Structure Was Pushing Budget Toward Low-Intent Placements
The root problem was structural. With one asset group per campaign and no clear segmentation by product category or purchase intent, Google had no framework for matching the right creative to the right audience at the right stage. The algorithm defaulted to the path of least resistance: broad Display and YouTube placements where impressions are cheap and the conversion bar is low.
Placement reports (pulled through scripts, since Google does not surface granular PMax placement data by default) showed a significant share of spend going to mobile app placements, auto-play YouTube pre-rolls on unrelated content, and low-quality Display inventory. These placements inflated impression volume while contributing almost nothing to actual purchases.
This is a common failure pattern in Performance Max. The algorithm is designed to find conversions wherever it can. When asset groups are undifferentiated, the algorithm treats the entire product catalog as one homogeneous offer and distributes accordingly. The signal quality feeding the algorithm matters more than the bidding strategy sitting on top.
The Missing Video Assets And What Google Auto-Generated Instead
Performance Max video asset best practices are clear: upload purpose-built video that communicates your value proposition, demonstrates your product, and matches the intent level of the audience you want to reach. When you do not upload video, Google auto-generates it. And Google's auto-generated video is, functionally, a slideshow of your static images with text overlays and stock transitions.
These auto-generated videos ran on YouTube placements, consuming real budget. They had no product demonstration, no narrative, no hook. Click-through rates were negligible. But because the algorithm counts view-through conversions on YouTube, the campaign reporting still attributed some value to these placements, masking the waste.
The auto-generated video problem is not a minor creative quibble. It is a structural budget leak. Google treats the presence of video as a signal that the campaign is eligible for YouTube inventory. If the video is poor, the algorithm still serves it. You get the spend without the performance. This is one of the most common Performance Max auto-generated video problems ecommerce brands face, and it is almost always invisible in default reporting.
Why The Learning Phase Kept Resetting After Routine Optimizations
Every time the team made changes, whether adjusting ROAS targets, swapping headlines, or pausing products, the campaign re-entered learning. With only one asset group per campaign, every edit affected the entire campaign's signal set. The algorithm never had a stable foundation to optimize from.
This created a frustrating loop: performance dips after a change, the team waits, performance partially recovers, another change is made, and the cycle repeats. The account was perpetually in a state of partial learning, never reaching the steady-state optimization that Performance Max is designed to deliver at scale.
The Intervention: Rebuilding PMax Asset Groups Around Audience Intent
Step 1: Auditing Placement And Asset-Level Performance Data
The first move was pulling granular data. Placement reports, asset performance ratings, and conversion path analysis were combined to build a picture of where budget was actually going and which creative assets were contributing to real conversions versus view-through noise.
This audit revealed that roughly a third of total PMax spend was going to placements that generated fewer than 2% of conversions. The auto-generated videos were the biggest offenders, but certain Display placements were also consuming budget with almost no return.
Step 2: Replacing Auto-Generated Video With Purpose-Built Creative
The brand produced a set of short-form videos, 15 and 30 seconds, tailored to specific product categories. These were not high-production commercials. They were straightforward product demonstrations, unboxing sequences, and benefit-focused clips shot on a phone with clean lighting and clear audio.
Google Ads Performance Max video requirements are not demanding from a production standpoint. What matters is relevance, clarity, and the presence of a clear value proposition in the first three seconds. The goal was to give the algorithm video that could actually convert, replacing the auto-generated slideshows that were consuming YouTube budget without earning clicks.
Each video was matched to a specific asset group (see next step), ensuring that the creative shown on YouTube placements was relevant to the product category being promoted.
Step 3: Restructuring Asset Groups By Product Category And Customer Intent Signal
This was the core of the rebuild. The single-asset-group-per-campaign structure was replaced with a segmented architecture: each campaign now contained multiple asset groups organized by product category, with creative, headlines, descriptions, and audience signals matched to the specific products in that group.
The pmax asset group strategy for ecommerce followed a clear logic:
- High-margin hero products got their own dedicated asset groups with the strongest creative and the most specific audience signals
- Mid-tier products were grouped by category with shared creative themes
- Lower-margin or long-tail products were grouped broadly, with tighter spend controls to prevent them from absorbing disproportionate budget
Each asset group received its own set of images, videos, headlines, and descriptions, all written and designed for the specific product category it represented. This gave the algorithm a much clearer signal about what to serve to whom.
Step 4: Setting Spend Controls And Audience Signals That Guided The Algorithm
Audience signals were layered into each asset group using a combination of first-party data through Customer Match and in-market segments relevant to each product category. These signals do not restrict targeting in Performance Max, but they give the algorithm a starting point that is far more productive than a cold start.
Spend controls were set at the campaign level to ensure hero products received priority budget, while lower-priority asset groups were capped to prevent runaway spend on low-margin items. The team also excluded specific placement categories (mobile app interstitials, for example) where prior data showed consistently poor performance.
The Results: What Changed And Why
Revenue Lift Without Increasing Total Budget
Within the first four weeks of the restructured campaigns exiting learning, total revenue from Google Ads increased meaningfully on the same monthly spend. The improvement came entirely from reallocation: budget that had been leaking into low-intent Display and YouTube placements was now flowing toward Shopping and Search placements where purchase intent was higher.
The brand did not need more budget. It needed its existing budget to stop subsidizing auto-generated video impressions on YouTube channels that had nothing to do with its products.
How ROAS Changed Once The Algorithm Had Better Creative Inputs
ROAS improved, but the more important shift was in the quality of conversions driving that number. Before the rebuild, the ROAS figure was partly inflated by view-through conversions from auto-generated video. After the rebuild, a larger share of reported conversions were click-through purchases with clear attribution, which translated directly into actual revenue growth.
CPA dropped as the algorithm spent less time exploring low-quality placements. The hero product asset groups, with their strong creative and specific audience signals, consistently outperformed the broader groups, validating the segmentation approach.
The Role Of The groas Engine In Continuous Optimization Post-Rebuild
Rebuilding the asset group architecture was the intervention. Keeping it optimized was the ongoing challenge. This is where groas changed the equation. The proprietary engine, trained on over $500 billion in profitable ad spend, monitored asset group performance continuously and made micro-adjustments to bidding, audience signals, and spend allocation around the clock.
For in-house teams and agencies, this level of continuous optimization is physically impossible. A human can check asset group performance once a day, maybe twice. The groas engine does not stop. In a DWY engagement, the engine handles the heavy lifting while a senior strategist works alongside the in-house team to make structural decisions. In a DFY engagement, a dedicated groas strategist owns the entire account end to end, including the creative strategy and landing page optimization that most agencies never touch.
The brand in this case transitioned from managing its own PMax campaigns to a collaborative model where its in-house marketer stayed in the driver's seat but had the groas engine and a senior strategist backing every decision. The result was not just a one-time fix, but a compounding improvement as the engine learned the account's specific conversion patterns.
The Lessons: What This Account Taught Us About PMax At Scale
Why Auto-Generated Assets Are A Performance Liability, Not A Convenience
Google's auto-generated video exists to ensure Performance Max can access YouTube inventory regardless of what the advertiser uploads. It is designed for Google's benefit, not yours. Every ecommerce brand running PMax without custom video is paying for YouTube placements with creative that will not convert. This is not a theoretical risk. It is a measurable budget leak.
Performance Max video asset best practices start with one rule: never let Google generate your video. Even a basic product demonstration shot on a smartphone will outperform an auto-generated slideshow, because it gives the viewer a reason to click.
How Asset Group Structure Determines Where Budget Actually Goes
Asset groups are not an organizational convenience. They are the primary mechanism through which Performance Max decides which creative to serve to which audience on which placement. A poorly structured asset group is an instruction to the algorithm to treat your entire catalog as a single undifferentiated offer. The algorithm will comply, and the results will be exactly as generic as the input.
The scaling roadblocks ecommerce brands miss are often structural rather than tactical. Increasing budget on a poorly structured PMax campaign does not scale performance. It scales waste.
When Performance Max Outperforms Search And When It Does Not
Performance Max is strongest when it has clear signals: well-structured asset groups, strong creative, specific audience signals, and a product feed that is properly optimized. Under those conditions, it can find incremental reach across placements that a Search-only campaign cannot access.
When those signals are weak, Performance Max defaults to the cheapest available inventory, which is almost always low-intent Display and auto-play YouTube. In those scenarios, a well-structured Search campaign will outperform PMax on every metric that matters.
What In-House Teams And Agencies Can Apply Today
The transferable lesson from this account is straightforward: Performance Max is not a set-it-and-forget-it campaign type. It requires deliberate architecture, purpose-built creative (especially video), and continuous monitoring of where budget is actually going.
For in-house teams running PMax, audit your asset groups now. If you have a single asset group per campaign, you are almost certainly leaking budget. If you have not uploaded custom video, Google is generating video for you and spending your money on it. Pull placement reports. Look at asset performance ratings. The data is there, even if Google does not surface it prominently.
For agencies managing multiple ecommerce accounts, the pmax asset group strategy described here scales across clients. groas gives agencies direct access to its proprietary engine through the DIY product, so media buyers can apply this level of optimization across their entire client book without adding headcount. Start your 7-day free trial and connect unlimited client accounts under one subscription.
For brands that do not have the in-house expertise or bandwidth to rebuild and maintain this level of PMax structure, groas offers a fully managed DFY service where a dedicated strategist owns your entire Google Ads account, including creative strategy, landing pages, and the continuous optimization that makes the difference between a one-time fix and compounding performance. There is no onboarding fee, no long-term contract, and the team is reachable on Slack or email around the clock. Apply today and find out what your account is leaving on the table.
Frequently Asked Questions
Why Does Google Auto-Generate Video For Performance Max Campaigns?
Google auto-generates video assets when an advertiser does not upload custom video to a Performance Max campaign. This ensures the campaign can access YouTube and video placement inventory. However, auto-generated videos are simple slideshows made from static images, brand colors, and headline text. They lack product demonstrations, narrative hooks, or meaningful calls to action. Because Google counts view-through conversions from these placements, the waste is often invisible in default reporting. The result is real budget spent on creative that rarely drives clicks or purchases. Uploading even basic purpose-built video eliminates this leak and gives the algorithm assets that can actually convert.
What Is The Best PMax Asset Group Strategy For Ecommerce?
The best pmax asset group strategy for ecommerce segments asset groups by product category and customer intent rather than lumping everything into a single group per campaign. High-margin hero products should get dedicated asset groups with tailored creative, headlines, and audience signals. Mid-tier products can be grouped by category, and lower-margin items should receive tighter spend controls. Each asset group needs its own images, videos, and copy matched to its specific products. This structure gives the algorithm clear signals about what creative to serve to which audience, preventing budget from defaulting to low-intent placements.
How Do I Know If My Performance Max Campaign Is Wasting Budget?
Pull placement reports using scripts or the Insights tab and look for disproportionate spend on mobile app placements, low-quality Display inventory, or auto-play YouTube pre-rolls on unrelated content. Check asset performance ratings for any assets rated "Low." Compare click-through conversions against view-through conversions. If a large share of reported value comes from view-throughs, your ROAS may be inflated. Rising CPAs alongside stable or declining revenue is another key warning sign that budget is flowing to the wrong placements.
What Are Google Ads Performance Max Video Requirements?
Google Ads Performance Max video requirements are not technically demanding. You need videos in landscape (16:9), square (1:1), or vertical (9:16) formats, ideally 15 or 30 seconds long. The content matters far more than production quality. A clear product demonstration, a benefit statement in the first three seconds, and a direct call to action will outperform any polished brand video that lacks those elements. The critical point is to upload custom video at all, because failing to do so means Google auto-generates video and spends your budget on it.
Why Does Performance Max Keep Re-Entering The Learning Phase?
Performance Max re-enters learning whenever significant changes are made to a campaign's asset groups, bidding strategy, audience signals, or creative. If you have only one asset group per campaign, every edit, no matter how small, affects the entire campaign's signal set. This triggers repeated learning phases that prevent the algorithm from reaching steady-state optimization. Segmenting asset groups isolates the impact of changes, so editing one group does not reset learning for the entire campaign.
Can Performance Max Replace Search Campaigns Entirely?
Not always. Performance Max outperforms Search when it has strong signals: well-structured asset groups, purpose-built creative, specific audience signals, and an optimized product feed. Under those conditions, it finds incremental reach across placements Search cannot access. When those signals are weak, PMax defaults to cheap, low-intent Display and YouTube inventory, and a well-structured Search campaign will outperform it on every metric that matters. The right approach is usually running both, with PMax handling broad discovery and Search capturing high-intent queries.
How Does groas Prevent Performance Max Budget Waste?
groas prevents PMax budget waste through its proprietary engine trained on over $500 billion in profitable ad spend, which monitors asset group performance and makes micro-adjustments to bidding, audience signals, and spend allocation continuously. A human strategist cannot check asset group performance more than once or twice a day. The groas engine never stops. In a DFY engagement, a dedicated strategist also owns creative strategy and landing page optimization, catching structural problems like auto-generated video waste before they compound.
Should I Upload Multiple Videos Per Asset Group?
Yes. Uploading multiple videos per asset group in different aspect ratios (landscape, square, and vertical) gives the algorithm options for different placements and screen sizes. Each video should be tailored to the specific product category in that asset group. Two to three variants per asset group, covering different angles like product demonstration, benefit highlight, and social proof, give the algorithm enough material to test and optimize without defaulting to auto-generated creative.
How Long Does It Take To See Results After Restructuring PMax Asset Groups?
Expect the restructured campaigns to spend roughly two to four weeks in the learning phase before performance data becomes reliable. After exiting learning, improvements in CPA and revenue attribution typically become visible within the first month. The compounding benefits of better signal quality and continuous optimization build over time. groas accelerates this timeline because its engine processes performance signals around the clock, shortening the gap between restructuring and measurable results compared to manual management.
What Is The Biggest Mistake Ecommerce Brands Make With Performance Max?
The biggest mistake is treating Performance Max as a set-and-forget campaign type. Brands launch with minimal asset group segmentation, skip custom video uploads, and assume the algorithm will figure everything out. The result is budget leaking into low-intent placements while the dashboard reports stable ROAS, masking declining revenue contribution. The fix is deliberate architecture: segmented asset groups, purpose-built creative, layered audience signals, and continuous monitoring of where budget actually goes.
.png)
