Google has introduced an A/B testing feature for its AI-driven ad tool, Performance Max (Pmax), allowing advertisers to assess changes like creative assets and landing pages. Announced in late October, the update will enable additional assets and dynamic landing page selection in a November beta. Some advertisers have accessed this feature through a whitelist program for over a month. Andy Goodwin of Brainlabs noted that A/B testing provides vital insights, countering Pmax's previously criticized "black-box" nature, which lacked transparency. Michelle Merklin from Tinuiti echoed similar concerns, stating that understanding campaign impact was challenging. The new testing shows real-time effects, unlike unreliable before-and-after methods. Despite the added controls, some advertisers question result reliability. Google responded to long-standing requests for more control, previously hinting at this feature 18 months ago but facing delays.

Now, advertisers can simultaneously run a base and experimental campaign, evenly distributing traffic across platforms like search and YouTube to determine which performs better. Merklin reported nearly 10 clients seeking access to test strategies for customer acquisition, smart bidding, audience targeting, and more. This offers advertisers clearer insights accounting for variables like seasonality. However, not all embrace the feature. An unnamed marketing agency executive said full agency adoption lags due to desired control over test elements. Unlike traditional tests, Pmax A/B tests within Pmax accounts can complicate isolating change impacts, raising concerns over transparency. Rain the Growth Agency’s past tests found Google reports tend to overstate performance, with director Stacia Fulginiti wary of adopting these experiments fully. A Google spokesperson clarified advertisers can define experiment traffic splits, ensuring balanced test conditions.