What Is Performance Max

In the Spanish-language Google Ads interface the campaign type is called Máximo rendimiento, although the industry almost always sticks to the original name. The difference from the classic campaign types lies in the starting point: in a Search campaign the advertiser picks keywords and decides where to compete, whereas here the advertiser hands over conversion goals, creative material and audience signals, and the system decides on which Google surface each impression is served.

The unit of work is the asset group (grupo de recursos): a set of headlines, descriptions, images, videos and logos bundled by theme or by audience. From that material Google assembles the specific formats each channel needs, from a text ad in Search to a video on YouTube. In e-commerce accounts the Merchant Center product feed is added, which makes the campaign the direct successor to the old smart Shopping campaigns.

Two things that are often confused should be kept apart. Performance Max is not a bidding technology but a campaign type; bidding is handled by Smart Bidding, which is also available in ordinary Search or Shopping campaigns. An account allows a maximum of 100 campaigns of this type, so the structure has a ceiling and needs planning.

Once live, the campaign competes in auctions on Search, Display, YouTube, Discover, Gmail and Maps, plus search partners. The split across those surfaces is decided by the system based on where it expects the most conversions towards the stated goal, and the advertiser cannot allocate budget channel by channel.

Within a single account, coexistence with other campaigns follows concrete rules. If a query matches an active exact match keyword in a Search campaign, that campaign takes priority over Performance Max. When no keyword matches exactly, the campaign with the higher Ad Rank is served. That second case explains most of the overlaps that show up with brand campaigns.

The historical black box criticism has lost much of its basis. The Performance Max search terms report provides data from March 2023 onwards, and the channel performance report, which breaks out impressions, clicks, cost, conversions and value per surface, covers periods after 6 June 2025. In January 2026 that channel segmentation also reached the Google Ads API, so it can be pulled into an in-house dashboard instead of being read by hand. What is still missing are distribution controls: traffic can be limited through campaign-level negative keywords, which act only on Search and Shopping inventory, and through brand exclusion lists, which additionally reach YouTube Search. Neither tool extends to Display, Discover or Gmail.

Three expensive decisions depend on this campaign type. The first is how much budget is taken out of Search and Shopping campaigns that are already profitable. Because Performance Max can cover the same queries, moving money without first checking the overlap shifts the internal split of the account long before it shifts total business volume.

The second is how the results are read. Brand traffic converts easily, because anyone searching the company name already knew the company. If Performance Max absorbs those queries, the campaign's return rises without any new demand entering, and the comparison with a generic campaign stops making sense. That is why common practice is to exclude the own brand from the campaign and measure it separately, in a manually controlled Search campaign.

The third decision concerns creative work. With no keyword bids left to adjust, the remaining optimisation lever is the material: which asset groups exist, what messages they carry and which audience signals accompany them. In e-commerce accounts feed quality is added to that, since it determines which queries a product can appear for at all.

• Exclude the own brand through brand exclusion lists or negative keywords and keep a separate brand Search campaign, so the two streams can be read apart.
• Record brand Search volume before and after launching the campaign: if it falls while Performance Max rises, traffic has shifted rather than grown.
• Build several asset groups by theme, margin or product line instead of a single one holding the whole catalogue, so channel and asset reports can be interpreted.
• Review the search terms report on a fixed schedule and move whatever is irrelevant into the campaign negative list.
• Keep the bidding strategy stable throughout the learning period and change only one parameter at a time.
• Maintain feed titles, attributes and images in e-commerce accounts, because much of the campaign's real targeting comes from there.

• Launching the campaign with poorly defined conversion goals, for instance counting a visit to a contact page as a sale.
• Judging performance after a few days and reacting to the noise of the learning period.
• Failing to exclude the brand and crediting the campaign with a return that comes from demand that already existed.
• Duplicating the same asset group across several campaigns and losing any way to tell which material works.
• Comparing CTR or CPM across channels in the performance report as if they were equivalent, when a video and a text ad are not consumed the same way.