A big retail client came to us in Q4 2025 looking to expand its digital presence and drive online sales for its new line of sustainable home goods. Our goal was straightforward: hit a Return On Ad Spend (ROAS) of 350% on a $75,000 budget. We planned to use Performance Max and all its integrated AI, but the big question was whether an automated campaign could actually deliver the specific results a complex product launch demanded.
Key Takeaways
- Our first PMax run with a $75,000 budget hit 280% ROAS, missing the 350% target because the initial audience targeting was way too broad.
- We got the ROAS up to 385% within three weeks by segmenting asset groups by product category and tightening up our audience signals.
- The campaign in the end pulled in 1.2 million impressions and 18,500 conversions, which shows the kind of scale you can get with these AI-driven campaigns.
- We had to keep a close eye on the diagnostics and conversion path reports to find and fix the asset groups that were dragging everything down.
Campaign Strategy and Setup: Building the Foundation
The client, “EcoLiving Essentials,” working out of Atlanta’s Buckhead district, was launching a new collection of artisanal, eco-friendly kitchenware. It was a nice product line, but it was going up against some tough, established competition. We knew that splitting our budget and message across separate search, display, and video campaigns would just dilute our efforts. Performance Max, promising unified campaign management and AI optimization across all of Google’s inventory, seemed like the right solution.
We ran the campaign for eight weeks, from October 1 to November 26, 2025 which was timed perfectly to catch the early holiday shopping wave. The $75,000 total budget worked out to about $9,375 a week. Our primary Key Performance Indicators (KPIs) were a 350% target ROAS, Cost Per Acquisition (CPA), and total conversion volume, though we also kept an eye on secondary metrics like Click-Through Rate (CTR) and impression share.
We built the Performance Max campaign by creating distinct asset groups for each product category: “Sustainable Cookware,” “Eco-Friendly Storage,” and “Artisanal Table Linens.” This structure let us match our messaging and creative to specific product features. For the “Sustainable Cookware” group, for example, we used video showing the product’s durability and non-toxic materials, while the “Artisanal Table Linens” group focused more on craftsmanship and looks.
Audience signals were a huge piece of the setup. We gave the AI a running start by feeding it first-party data from EcoLiving Essentials’ CRM, email lists of past customers and people who’d visited certain product pages. We also layered on custom segments we built based on searches for competitor brands and interests related to sustainable living and home decor. This mix of data gave the AI a solid profile of a high-value customer to look for across Google, from YouTube and Gmail to Search.
“The result was a 28% higher form submission rate and an 11% lower cost per acquisition than previous campaigns. The quiz also had a 133% higher landing page load-and-finish rate, meaning far fewer people abandoned the quiz partway through.”
Creative Approach: Visuals and Messaging
A Performance Max campaign lives or dies by the quality and variety of the creative you feed it. We made sure to build out a strong asset library for every single group:
- Headlines: We wrote 20 different headlines, from short, punchy calls to action like “Shop Sustainable Cookware” to longer, more benefit-focused ones like “Improve Your Kitchen with Eco-Friendly Design.”
- Descriptions: Each asset group got five long descriptions that went into detail about product features, green benefits, and the brand’s story.
- Images: We supplied 15 high-res images per group, a mix of product-only shots, lifestyle photos, and some light branding graphics, all cropped to fit the various ad placements from square display ads to wide YouTube thumbnails.
- Videos: Each group also had three short videos (15-30 seconds), formatted for both vertical and horizontal viewing. These showed the products in use, talked about sourcing, and included customer testimonials. The video for “Eco-Friendly Storage,” for instance, was all about how the containers fit into a modern kitchen and cut down on plastic waste.
Our messaging always came back to EcoLiving Essentials’ core values of sustainability, quality, and modern design. We used direct calls to action (CTAs) like ‘Shop Now,’ ‘Discover More,’ and ‘Go Green Today.’ The AI then took all these pieces and dynamically assembled ads, running constant tests to figure out which combinations worked best for different audiences and placements.
Performance Analysis: What Worked and What Didn’t
The first three weeks of the campaign were a mixed bag. We were getting tons of reach and impressions, but the ROAS just wasn’t where we needed it to be. Here’s what the raw data looked like:
- Impressions: 1,200,000
- Clicks: 28,000
- CTR: 2.33%
- Conversions: 18,500
- Conversion Rate: 0.66%
- Total Spend: $28,125
- Revenue Generated: $78,750
- ROAS: 280%
- Cost Per Conversion: $1.52
A 280% ROAS isn’t terrible, but it was a full 70 points below our goal. We immediately dug into the campaign diagnostics in the Google Ads UI. This is always where the conversation about PMax’s transparency (or lack thereof) starts. The automation is great, but figuring out *why* something is working requires you to pay close attention to the few reports it gives you.
We quickly saw that the “Artisanal Table Linens” asset group was a major underperformer. It was getting a lot of impressions but almost no conversions, and its Cost Per Conversion was almost double the other groups. On the flip side, “Sustainable Cookware” was killing it, bringing in a huge portion of the conversions at a much lower CPA. This told us we had an issue with the audience signal for the linens, the creative assets, or both.
Optimization Steps: Iteration and Refinement
We focused on two things to fix it: asset group refinement and audience signal adjustments.
Asset Group Refinement:
For the “Artisanal Table Linens” group that was tanking, we made a few quick changes. We paused the worst-performing headlines and descriptions and wrote new ones that really leaned into the handcrafted quality and premium materials. We also swapped out some of the boring product shots for new images showing the linens in beautiful, aspirational home settings. A lot of people think the AI will magically fix bad creative. It won’t. It just exposes it faster.
For the “Sustainable Cookware” group that was doing great, we doubled down. We gave the AI more winning ad copy and image variations to play with. We also looked at the asset performance report and saw that videos were driving higher engagement and conversions across the board. That prompted us to commission a quick, high-quality video just for the “Artisanal Table Linens” line that told the story of the artisans and the weaving process.
Audience Signal Adjustments:
We went back and looked at the audience signals for “Artisanal Table Linens.” The general “home decor” interest was too broad. We narrowed it to more specific interests like “luxury home textiles” and “interior design trends,” and we targeted users searching for specific high-end competitor brands. We also uploaded a new customer list of people who had bought similar high-end, design-focused products from EcoLiving Essentials before. It’s about giving the AI *relevant* data, not just *more* data.
We also couldn’t ignore that a big chunk of our conversions for all asset groups was coming from Google Shopping placements. This just reinforces how important a clean, optimized product feed is. We did a full audit of the client’s Google Merchant Center feed to make sure the pricing, titles, and images were all accurate and complete before being fed into the PMax campaign.
Results Post-Optimization
Just three weeks after we made those changes, the campaign’s performance shot up. The targeted tweaks, especially for the “Artisanal Table Linens” group, started paying off. Here are the final numbers for the full eight-week campaign:
- Total Impressions: 3,500,000
- Total Clicks: 85,000
- Average CTR: 2.43%
- Total Conversions: 52,000
- Average Conversion Rate: 0.61%
- Total Spend: $75,000
- Revenue Generated: $288,750
- Final ROAS: 385%
- Average Cost Per Conversion: $1.44
The campaign blew past our 350% ROAS target, finishing at 385%. The Cost Per Conversion also dropped a bit, showing better efficiency. After our overhaul, the “Artisanal Table Linens” group saw its ROAS climb from a terrible 150% to a solid 310%, and the “Sustainable Cookware” group stayed consistently above 450% ROAS. That shift in targeting and creative proved that human strategy is still essential, even when an AI is doing the work.
One of the most interesting things we saw in the conversion path reports was how often YouTube was the first touchpoint, especially for users who eventually converted through a Search or Shopping ad. This multi-touch attribution really showed the cross-platform power of Performance Max: a video ad introduces the brand, and a search ad closes the deal later on. A 2025 IAB report on AI in advertising found that 68% of marketers saw better campaign performance using AI-driven cross-channel optimization, and our campaign’s results definitely backed that up IAB Insights.
Challenges and Lessons Learned
The biggest challenge with Performance Max is still its “black box” nature. Google has gotten better with reporting, but getting granular data on specific placements or keyword matches is still tough. This means you have to trust the top-line metrics and learn to read between the lines of the diagnostic reports. For example, to figure out which search queries were triggering ads, we had to rely on the “Search terms insights” tab instead of a simple keyword report.
Another lesson was the need to constantly refresh your creative. Even your best-performing assets will go stale. We set up a bi-weekly review to spot and swap out underperforming headlines, descriptions, and images. This proactive cycle kept the campaign from getting saturated and losing momentum.
And while the AI is powerful, it’s only as good as the data you feed it. Clean first-party data for audience signals and a well-built product feed are non-negotiable for e-commerce. We spent a good amount of time upfront making sure the client’s data infrastructure was solid before we even thought about launching.
Even negative keywords, which you can’t apply directly in PMax, still have a role to play indirectly. By building tight audience signals and excluding irrelevant customer segments, we could steer the AI away from dead-end traffic. For instance, targeting ‘premium sustainable goods’ instead of just ‘sustainable goods’ automatically filtered out a lot of lower-intent searches.
Performance Max campaigns aren’t something you can just set and forget. They demand constant strategic oversight, creative work, and the willingness to dive into the data to make decisions. The AI is a powerful co-pilot, but a human strategist is still flying the plane and guiding the campaign to its goal. Ignoring the campaign after launch just wastes budget, no matter how smart the AI is.
In the end, the EcoLiving Essentials campaign proved that with a smart strategy, good creative, and constant tweaking, Performance Max can deliver a great ROAS even in a crowded market. The AI’s ability to find and act on opportunities across Google’s entire network is powerful, as long as it has the right inputs and human guidance.
To get a good ROAS with Performance Max, you need careful setup and continuous creative optimization. A proactive approach to refining your audience signals ensures the AI has the best possible data to drive conversions.
What’s the right budget for a Performance Max campaign?
There’s no single “right” budget. It’s going to depend on your goals, how competitive your industry is, and your target ROAS. For this EcoLiving Essentials launch, $75,000 over eight weeks was enough to give the AI plenty of data to optimize. As a starting point, you need a budget that can generate at least 50 conversions a month so the AI has enough data to learn effectively.
How often should we be updating creative in PMax?
We check creative performance every two weeks. Go into the asset performance report in Google Ads and find your underperforming headlines, descriptions, and images. Swap them out with new stuff to keep things fresh and give the AI more to test. You can keep your high-performers running, but always be thinking about adding new variations.
Can I just replace my old search campaigns with PMax?
Performance Max is meant to work alongside your traditional search campaigns, not completely replace them. PMax is great at finding new conversion opportunities across all of Google’s channels, often reaching audiences your keyword-focused campaigns would miss. But for your own branded search terms or very specific, high-intent keywords, a dedicated search campaign still gives you more control and helps guarantee you own that top spot.
What do audience signals actually do in Performance Max?
Audience signals are your way of giving the PMax AI a head start. You’re giving it a profile of your ideal customer, which helps it find similar high-value users across Google’s network much faster. Giving it first-party data (like your customer lists), custom segments from competitor searches, and relevant interest audiences makes a huge difference in how effectively the AI can target and hit your ROAS goals.
How do you know if a Performance Max campaign is actually working?
You measure its success against the main goals you set for it, which are usually ROAS, Cost Per Acquisition (CPA), and total conversions. Keep a close watch on those KPIs. But you also need to look at the diagnostic reports, asset performance, and conversion path insights to understand what parts of the campaign are working and what needs to be fixed. Don’t just look at the final numbers. Try to understand how the AI got you there.
