Performance Max has come a long way since its rocky launch. Many advertisers once dismissed it as a half-baked product, but Google has spent the past 18 months fixing real issues around transparency and control. If you wrote Performance Max off before, it’s time to take another look.
Mike Ryan, head of ecommerce insights at Smarter Ecommerce, explained why at the latest SMX Next.
Taking a fresh look at Performance Max
Performance Max traces its roots to Smart Shopping campaigns, which Google rolled out with red carpet fanfare at Google Marketing Live in 2019.
Even then, industry experts warned that transparency and control would become serious issues. They were right — and only now has Google begun to address those concerns openly.
Smart Shopping marked the low point of black-box advertising in Google Ads, at least for ecommerce. It stripped away nearly every control advertisers relied on in Standard Shopping:
- Promotional controls.
- Modifiers.
- Negative keywords.
- Search terms reporting.
- Placement reporting.
- Channel visibility.
Over the past 18 months, Performance Max has brought most of that functionality back, either partially or in full.
Understanding Performance Max search terms
Search terms are a core signal for understanding the traffic you’re actually buying. In Performance Max, most spend typically flows to the search network, which makes search term reporting essential for meaningful optimization.
Google even introduced a Performance Max match type — something few of us ever expected to see. That’s a big deal. It delivers properly reportable data that works with the API, should be scriptable, and finally includes cost and time dimensions that were completely missing before.
Search term insights vs. campaign search term view
Google’s first move to crack open the black box was search term insights. These insights group queries into search categories — essentially prebuilt n-grams — that roll up data at a mid-level and automatically account for typos, misspellings, and variants.
The problem? The metrics are thin. There’s no cost data, which means no CPC, no ROAS, and no real way to evaluate performance.
The real breakthrough is the new campaign-level search term view, now available in both the API and the UI.
Historically, search term reporting lived at the ad group level. Since Performance Max doesn’t use ad groups, that data had nowhere to go.
Google fixed this by anchoring search terms at the campaign level instead. The result is access to far more segments and metrics — and, finally, proper reporting we can actually use.
The main limitation: this data is available only at the search network level, without separating search from shopping. That means a single search term may reflect blended performance from both formats, rather than a clean view of how each one performed.
Search theme reporting
Search themes act as a form of positive targeting in Performance Max. You can evaluate how they’re performing through the search term insights report, which includes a Source column showing whether traffic came from your URLs, your assets, or the search themes you provided.
By totaling conversion value and conversions, you can see whether your search themes are actually driving results — or just sitting idle.
There’s more good news ahead. Google appears to be working on bringing Dynamic Search Ads and AI Max reports into Performance Max. That would unlock visibility into headlines, landing pages, and the search terms triggering ads.
Search term controls and optimization
Negative keywords
Negative keywords are now fully supported in Performance Max. At launch, Google capped campaigns at 100 negatives, offered no API access, and blocked negative keyword lists—clearly positioning the feature for brand safety, not performance.
That’s changed. Negative keywords now work with the API, support shared lists, and give advertisers real control over performance.
These negatives apply across the entire search network, including both search and shopping. Brand exclusions are the exception — you can choose to apply those only to search campaigns if needed.
Brand exclusions
Performance Max doesn’t separate brand from generic traffic, and it often favors brand queries because they’re high intent and tend to perform well. Brand exclusions exist, but they can be leaky, with some brand traffic still slipping through. If you need strict control, negative keywords are the more reliable option.
Also, Performance Max — and AI Max — may aggressively bid on competitor terms. That makes brand and competitor exclusions important tools for protecting spend and shaping intent.
Optimization strategy
Here’s a simple heuristic for spotting search terms that need attention:
- Calculate the average number of clicks it takes to generate a conversion.
- Identify search terms with more clicks than that average but zero conversions.
Those terms have had a fair chance to perform and didn’t. They’re strong candidates for negative keywords.
That said, don’t overcorrect.
Long-tail dynamics mean a search term that doesn’t convert this month may matter next month. You’re also working with a finite set of negative keywords, so use them deliberately and prioritize the highest-impact exclusions.
Modern optimization approaches
It’s not 2018 anymore — you shouldn’t spend hours manually reviewing search terms. Automate the work instead.
Use the API for high-volume accounts, scripts for medium volume, and automated reports from the Report Editor for smaller accounts (though it still doesn’t support Performance Max).
Layer in AI for semantic review to flag irrelevant terms based on meaning and intent, then step in only for final approval. Search term reporting can be tedious, but with Google’s prebuilt n-grams and modern AI tools, there’s a smarter way to handle it.
Channels and placements reporting
Channel performance report
The channel performance report — not just for Performance Max — breaks performance out by network, including Discover, Display, Gmail, and more. It’s useful for channel visibility and understanding view-through versus click-through conversions, as well as how feed-based delivery compares to asset-driven performance.
The report includes a Sankey diagram, but it isn’t especially intuitive. The labeling is confusing and takes some decoding:
- Search Network: Feed-based equals Shopping ads; asset-based equals RSAs and DSAs.
- Display Network: Feed-based equals dynamic remarketing; asset-based equals responsive display ads.
Google also announced that Search Partner Network data is coming, which should add another layer of useful performance visibility.
Channel and placement controls
Unlike Demand Gen, where you can choose exactly which channels to run on, Performance Max doesn’t give you that control. You can try to influence the channel mix through your ROAS target and budget, but it’s a blunt instrument — and a slippery one at best.
Placement exclusions
The strongest control you have is excluding specific placements. Placement data is now available through the API — limited to impressions and date segments — and can also be reviewed in the Report Editor. Use this data alongside the content suitability view to spot questionable domains and spammy placements.
For YouTube, pay close attention to political and children’s content. If a placement feels irrelevant or unsafe for your brand, there’s a good chance it isn’t driving meaningful performance either.
Tools for placement review
If you run into YouTube videos in languages you don’t speak, use Google Sheets’ built-in GOOGLETRANSLATE function. It’s faster and more reliable than AI for quick translation.
You can also use AI-powered formulas in Sheets to do semantic triage on placements, not just search terms. These tools are just formulas, which means this kind of analysis is accessible to anyone.
Search Partner Network
Unfortunately, there’s no way to opt out of the Search Partner Network in Performance Max. You can exclude individual search partners, but there are limits.
Prioritize exclusions based on how questionable the placement looks and how much volume it’s receiving. Also note that Google-owned properties like YouTube and Gmail can’t be excluded.
Based on Standard Shopping data, the Search Partner Network consistently performs meaningfully worse than the Google Search Network. Excluding poor performers is recommended.
Device reporting and targeting
Creating a device report is easy — just add device as a segment in the “when and where ads showed” view. The tricky part is making decisions.
Device analysis
For deeper insight, dig into item-level performance in the Report Editor. Add device as a segment alongside item ID and product titles to see how individual products behave across devices. Also, compare competitor performance by device — you may spot meaningful differences that inform your strategy.
For example, you may perform far better on desktop than on mobile compared to competitors like Amazon, signaling either an opportunity or a risk.
Device targeting considerations
Device targeting is available in Performance Max and is easy to use, much like channel targeting in Demand Gen. But when you split campaigns by device, you also split your conversion data and volume—and that can hurt results.
Before you separate campaigns by device, consider:
- How competition differs by device
- Performance at the item and retail category level
- The impact on overall data volume
Performance Max performs best with more data. Campaigns with low monthly conversion volume often miss their targets and rarely stay on pace. As more data flows through a campaign, Performance Max gets better at hitting goals and less likely to fall short.
Any gains from splitting by device can disappear if the algorithm doesn’t have enough data to learn. Only split when both resulting campaigns have enough volume to support effective machine learning.
Conclusion
Performance Max has changed dramatically since launch. With search term reporting, negative keywords, channel visibility, placement controls, and device targeting now available, advertisers have far more transparency and control than ever before.
It’s still not perfect — channel targeting limits and data fragmentation remain — but Performance Max is fundamentally different and far more manageable.
Success comes down to knowing what data you have, how to access it efficiently using modern tools like AI and automation, and when to apply controls based on performance insights and data volume needs.
Watch: PMax reporting for ecommerce: What Google is (and isn’t) showing you
Search Engine Land is owned by Semrush. We remain committed to providing high-quality coverage of marketing topics. Unless otherwise noted, this page’s content was written by either an employee or a paid contractor of Semrush Inc.
