As ad platforms like Google automate bid management, an agency's value is no longer in manual "button pushing." The new competitive edge is the ability to feed the platform's AI with superior client data and insights. Agencies that cannot access and leverage this data will struggle to demonstrate value.

Unlike Netflix, which struggles with attribution using clean rooms and IP matching, ChatGPT's ad platform can leverage direct clicks. This allows for high-fidelity measurement similar to Meta's CAPI and pixel, providing advertisers with much clearer, less probabilistic attribution for their ad spend.

A novel way to measure ad effectiveness in LLMs is "attention shift"—analyzing how much an ad pivots the conversation's topic toward the brand. This metric, derived from vector analysis of messages before and after an ad, captures influence beyond traditional clicks or impressions, reflecting deeper engagement.

Unlike Meta or Google, OpenAI's early ad offering for ChatGPT will not provide detailed attribution data or conversion tracking. Advertisers will only receive high-level metrics like impressions and clicks, a significant step back from the granular performance measurement they are accustomed to.

AI models for campaign creation are only as good as the data they ingest. Inaccurate or siloed data on accounts, contacts, and ad performance prevents AI from developing optimal strategies, rendering the technology ineffective for scalable, high-quality output.

To effectively sell ads, OpenAI must provide advertisers with targeting tools and performance data. This will inadvertently open up a treasure trove of analytics for all marketers, offering the first real glimpse into user behavior, popular topics, and prompt trends within ChatGPT.

OpenAI's initial ad offering is intentionally basic (CPM-based, low targeting) to gather data and advertiser feedback. This MVP approach is necessary to build the foundation for a more sophisticated, conversion-optimized platform like Meta's, even if it seems underdeveloped at first.

Despite platform fragmentation, Digitas's CEO argues the job of advertising is fundamentally the same. For a data-driven "quant," the North Star has always been whether an action drove sales. The challenge isn't new complexity, but rather marketers clinging to outdated, unmeasurable goals like "setting culture."

ChatGPT's ad platform launched with a simple CPM model and limited targeting, similar to Netflix's. This isn't a sign of weakness but a strategic necessity. To build a sophisticated, conversion-optimized ad system, a platform must first launch a "primitive MVP" to bootstrap the very conversion data required for advanced targeting.