A new product from Adobe Advertising lets brands use their own analytics data to train AI for campaign optimization. Marketers gain more control and transparency over ad performance.
Adobe Advertising has launched its Custom Algorithms product, giving marketers the ability to optimize programmatic campaigns using their own first-party data directly from Adobe Analytics. This move is designed to offer brands more control and transparency over how AI models bid and measure ad performance, a shift that could impact how publishers and advertisers approach campaign management.
The Custom Algorithms product, now out of beta, marks a significant change from the earlier TubeMogul era, according to Erwin Castellanos, GM of Adobe Advertising. The platform now features agentic plug-ins, advanced campaign controls, and performance-based measurement tools, aligning Adobe with other major ad platforms that have introduced AI-driven campaign solutions.
Previously, Adobe’s DSP accessed Adobe Analytics data through an API, similar to third-party DSPs. With the new integration, brands using Custom Algorithms can leverage raw analytics data, which Castellanos said provides a more complete view of the customer journey, including app and web traffic, media engagement, and first-party identity data that is typically not shared with DSPs.
Greg Collison, head of product and design at Adobe Advertising, noted that while other platforms like Google’s Performance Max and Meta’s Advantage+ Shopping Campaigns use their own first-party data for optimization, Adobe’s approach relies solely on the advertiser’s data or, in some cases, data from trusted partners such as retail media networks. Adobe does not own the media inventory, and Castellanos argued that using a brand’s own data is more cost-effective than relying on third-party or walled garden data, referencing the higher costs associated with platforms like The Trade Desk.
However, the effectiveness of Custom Algorithms depends on the volume and quality of a brand’s first-party data. Sectors like travel, ecommerce, and subscription services are likely to benefit most, while CPG brands may see less advantage due to limited direct customer data. To address this, Adobe is helping marketers define "High Value Actions"-metrics based on behaviors beyond purchases, such as video views, newsletter signups, or engagement with marketing content. These signals can be used to train bidding models even when transaction data is scarce.
Brands can also incorporate surveys and upper-funnel metrics like brand recognition or sentiment to further inform campaign optimization. Castellanos emphasized that both walled gardens and independent DSPs would prefer access to this level of data, but Adobe’s integration offers a unique approach for advertisers seeking more transparency and control.
For those interested in the broader implications of automation and agentic advertising, a related discussion on the risks of repeating programmatic’s intermediary challenges can be found in this analysis of automation in digital ad sales.
