Barcelona | Broadsign presented the industry’s first AI-based workflow for creative categorization and approval of pDooH campaigns on the sidelines of ISE. The new DooH feature is expected to be available from April.
Broadsign stopped exhibiting at ISE two years ago. Nevertheless, the Canadians were on site in Barcelona with a large management team to host its annual global user conference, Broadsign Connect, in the city center alongside ISE. There, the undisputed DooH market leader introduced new AI-powered features to automate approval workflows for ad creatives, speeding up the processing for programmatic DooH (pDooH) campaigns.
This new AI Assistant is a patent-pending tool designed to significantly reduce the time media owners spend on repetitive tasks like reviewing, categorizing, and approving campaign creatives. Large DooH media owners manually process tens of thousands of programmatic campaign creatives each month from connected demand-side platforms (DSPs). Currently, the volume of creatives requiring review doubles every year.
How AI-assisted approval workflows work at Broadsign
When campaign creatives are submitted to the Broadsign Supply-Side Platform (SSP), the Broadsign AI Assistant sends the network operator emails with category and approval suggestions for each ad. These recommendations are based on detailed creative analysis and insights from the media owner’s existing inventory taxonomy to enable competitive differentiation across channels.
The technology analyzes creatives for quality factors like aspect ratio, resolution, and inappropriate language, as well as objects that help determine the best category in the DooH network operator’s taxonomy—for example, identifying a car for the automotive category. Based on this analysis, the AI assistant suggests approvals and categorizations, which can be accepted, rejected, or forwarded for further review.
By automating categorization, the Broadsign AI Assistant aims to reduce classification errors and to quickly detect sensitive content that may compromise brand safety. For example, it can flag alcohol ads near schools or political ads displayed next to polling stations, ensuring compliance with local laws.
AI training with DSP bids received by Broadsign customers
Broadsign fine-tuned and trained the large language model (LLM) behind the AI Assistant using two years of categorization data aggregated from DSP bids received by Broadsign customers. Once live and in use, it will continue to learn and improve its recommendations based on each media owner’s specific categorization and approval practices.
The Broadsign AI Assistant is expected to launch in early Q2 2025 and will be available to customers using Broadsign SSP and Header Bidder.
