Google Performance Max launched in late 2021 and spent most of 2022 and 2023 as the campaign type that performance marketers complained about. No keyword visibility, no placement reporting, brand searches cannibalising Search campaigns, and search term reports that hid most queries. By 2026, Performance Max has matured significantly. Search themes were rolled out widely in 2024. Asset group reporting improved. Brand exclusions became more granular. Account-level negative keywords work properly now. The campaign type that frustrated agencies in 2022 is genuinely useful in 2026, if you set it up right.
This guide covers what we have learned running Performance Max across over 200 accounts in India, the UK, the USA and APAC markets through 2025 and 2026. What works, what to avoid, and how to think about Performance Max in the context of a broader Google Ads strategy.
What Performance Max actually is
Performance Max runs a single campaign that places ads across Search, Display, YouTube, Discover, Gmail and Maps. The advertiser provides asset groups (text, images, video, logos) and conversion goals. Google’s Smart Bidding decides where to show the ads, to whom, when, and at what price. The advertiser has limited control over those decisions.
That description has not changed. What has changed is how transparent Google has made the optimisation. The 2024 to 2026 updates added asset group-level reporting, audience signal performance, search themes that work alongside the algorithm’s own query discovery, and channel-level performance data. You can now see which channel within the campaign is driving conversions and adjust signals accordingly.
Asset group strategy
Asset groups are the building blocks of Performance Max. Each asset group has its own audience signal, listing group (for ecommerce), and creative assets. A campaign can hold up to 100 asset groups, but most accounts work best with 5 to 15.
The mistake we see most often is treating asset groups like ad groups in a Search campaign, splitting them by keyword theme. Performance Max does not respect that structure because the campaign does not run on keywords. We organise asset groups by audience intent instead: prospecting vs retargeting, new customer vs existing customer, brand awareness vs direct response, or for ecommerce, by product collection and margin tier.
For a D2C apparel brand we run, asset groups are split as: New customers seeing the brand for the first time (prospecting signal: lookalikes off purchasers), Existing customers we want to retain (retargeting signal: site visitors in last 30 days, past purchasers), Specific high-margin collections that need dedicated creative (signal: collection page visitors), and Seasonal promo campaigns during sales events. Each group gets distinct creative tuned to that audience.
Search themes done properly
Search themes were Google’s response to advertisers complaining they had no keyword input in Performance Max. They are essentially a list of phrases you tell Google “these are queries we want to compete on”. Google blends search themes with its own query discovery, so they are guidance not strict targeting.
Best practice in 2026 is to add 5 to 15 search themes per asset group, focused on intent rather than exact phrases. “Buy women running shoes online India” is a useful theme. “Running shoes” is too broad. We typically pull search themes from the top 25 converting queries in existing Search campaigns and add brand and competitor themes selectively.
Do not list more than 25 search themes per asset group. The algorithm starts ignoring them and the constraint hurts performance.
The brand search question
By default, Performance Max bids on your brand terms. For most accounts, this is a problem. The campaign reports inflated conversions because brand searches were already going to convert. CPAs look better than they should. Budget gets spent on traffic you would have won for free or cheaply through brand Search campaigns.
We add brand exclusions to every Performance Max campaign by default. Google added account-level brand exclusion lists in 2024. We use them. Brand Search runs as a separate dedicated campaign with manual CPC bidding, where ROAS is typically 15x or higher and we want maximum impression share without paying Performance Max prices.
Audience signals that actually move performance
Audience signals tell Performance Max who you want to reach. They are not hard targeting, just hints. The algorithm uses them as a starting point and expands outwards.
For prospecting asset groups, we layer custom audiences (people who searched recent keywords related to the product) plus customer match lists (lookalikes off past purchasers) plus in-market segments. Three layers gives the algorithm enough signal to find similar users while keeping the seed audience tight.
For retargeting asset groups, we use site visitors in the last 30 days, cart abandoners, and category page visitors as the audience signal. Google still expands beyond these, but the seed weights conversions toward warm prospects.
Creative assets at scale
Performance Max wants volume of creative. The recommended minimum is 15 images, 5 videos, 5 logos and 5 headlines per asset group, plus long headlines and descriptions. Most accounts under-supply creative, which means Google has fewer combinations to test, which limits performance.
We produce creative in batches of 30 to 50 image variants per asset group, 3 to 6 video variants, and full text asset coverage. Most variants get exported in all required ratios (1:1, 4:5, 9:16, 16:9) so the algorithm has placement flexibility. Video variants run 6 to 30 seconds, hook-led, with captions burned in for sound-off viewing.
Asset performance ratings (Low, Good, Best) update weekly. Replace any Low-rated asset within 2 weeks of receiving the rating. Keep Best assets running and use them as the template for the next creative batch.
When to use Performance Max and when not to
Performance Max works well for ecommerce with a healthy product feed and clear conversion event. It works well for lead generation with sufficient conversion volume (30 plus conversions per week per asset group). It works well for app installs.
It does not work well for: brands with low conversion volume (under 20 weekly conversions), brands with long sales cycles where the conversion event sits weeks after the click, or brands where keyword control matters more than channel breadth. For these accounts, stick with Search campaigns and add Performance Max only when the data volume justifies the trade-off.
Measurement and reporting
The native Performance Max reporting still hides more than we would like. We pull complete data through the Google Ads API into BigQuery and visualise in Looker Studio. Channel split (Search vs Display vs YouTube), asset group performance, search themes vs algorithmic queries, and audience signal effectiveness all become visible.
For ecommerce, we also overlay shopping reports to see product-level performance within Performance Max. Top sellers get more creative investment. Slow movers get pulled from feeds if they drain spend without converting.
What to expect from us
We run Performance Max as one part of a broader paid search strategy, not the whole strategy. Brand Search stays separate. High-intent commercial keywords stay in Search campaigns with manual bidding. Performance Max handles prospecting, broad commercial intent, and feed-driven shopping discovery. The split typically lands at 40 percent budget in Performance Max, 35 percent in Search, 15 percent in YouTube and Demand Gen, and 10 percent in remarketing.
If your account is over-relying on Performance Max (70 percent plus of budget), it is probably leaving performance on the table. If it is under-using Performance Max (under 20 percent), it is probably missing reach. Get the split right and the campaign type earns its place in the account structure.
