Michael McGoldrick: signal architecture becomes the new competitive edge in AI-run GTM

August 16, 2026 · thedrum.com

Michael McGoldrick, writing in The Drum, argues that AI agents entering go-to-market teams inherit a signal architecture problem that humans have quietly managed for years. He points to Gartner's forecast that AI agents will intermediate $15tn in B2B purchases by 2028, spanning supplier identification through order execution without a human buyer ever navigating a vendor's website. McGoldrick's argument centers on how B2B marketing teams have folded different signal types, website engagement, review site activity, bidstream intent, publisher network data, into a single dashboard or score, even though a prospect reading a comparison guide and a prospect downloading a whitepaper directly from a vendor represent different evidentiary categories. Human reps and marketers currently apply judgment to these blended signals, discounting a spike in anonymous intent data while trusting a return visit from a known account.

Budget allocation and outbound sequencing decisions that a human would normally sense-check today get executed by an AI agent with no equivalent instinct, McGoldrick writes, compounding the cost of a single misjudged signal across every account the agent touches, at whatever speed and scale it operates. The competitive edge, McGoldrick argues, is moving away from raw dataset size and toward a deliberate hierarchy of buyer intelligence: contact-level engagement with owned content tied to an identifiable account carries more weight than an inferred topic surge pulled from bidstream data with no buyer attached. Agencies and advertisers building AI-driven demand generation should audit which signals in their stack are observed and which are inferred before handing decision authority to an agent.