Last week, BCG released data that nobody wants to talk about: CMOs claim massive progress on agentic AI transformation. When researchers dug into what actually changed, they found almost nothing.
The gap isn't between vision and execution. It's between what leadership presents to the board and what's actually happening on the ground. It's between hype and infrastructure. And it's about to get a lot more painful.
The Funnel No Longer Exists
Here's what changed in the last six months that most marketing teams haven't adapted to yet.
An agent doesn't experience your brand the way humans do. When you or I shop, we see an ad, get curious, click through, browse, consider, add to cart. Three to five touchpoints. Multiple moments where emotion and creative do the heavy lifting.
An agent collapses that into seconds. Intent - research - comparison - purchase. No pauses. No emotional persuasion. No brand experience.
The IAB Australia Future of Search Working Group mapped this out in detail last week. Their conclusion was blunt: discovery, consideration, and purchase—the three stages that justified search advertising budgets for two decades—can now happen in a single sequence with zero human clicks.
Most CMOs haven't told their teams this yet.
Discoverability Killed Brand Awareness
When humans decide what to buy, emotion matters. A logo, a campaign, a story—these shape decisions. That's why brand awareness was worth 100 billion a year.
When agents decide, it doesn't. Agents read your product data, compare specs, check your credentials, and move on. They don't remember your Super Bowl spot. They don't feel your brand story. They evaluate factually or they don't evaluate you at all.
The report calls this shift from "awareness" to "discoverability." A brand that isn't machine-readable—that doesn't have clean, accessible product data and credentials—is invisible to agents regardless of how much you spend on creative.
Spend that 10 million Super Bowl budget now and see what happens. Your agent can't see it. It's not a marketing problem anymore. It's an infrastructure problem.
The Trust Collapse Is Already Here
Here's the number that should wake up every CMO in a board call: only 14% of Australian consumers trust organizations to use AI responsibly.
Yet 60% expect to use agentic AI daily within a year.
That's not confidence. That's capitulation. Consumers are ready to delegate to agents but they don't trust the companies building those systems.
When asked what they're most concerned about, the top two responses were loss of human contact (61%) and misuse of personal data (56%). And the kicker: 66% say they need the ability to manually override or stop an automated action.
If you're building an agent that can't be overridden by the user, you're building a trust liability. Most CMOs haven't started addressing this. The ones running agentic pilots are treating it as a tech problem. It's not. It's a brand problem.
The Infrastructure Is Still Infantile
Google's pushing Universal Commerce Protocol. Microsoft's rolling out NLWeb. Anthropic is evolving Model Context Protocol. All of these are less than six months old in practical deployment.
When you have four competing protocol layers all "solving" similar problems and none of them are standardized yet, you don't have infrastructure. You have chaos wearing a tech conference badge.
The report describes the agentic stack as "still in its infancy." That's generous. It's premature infancy. The babies haven't learned to walk.
Most enterprise marketing teams are building proofs of concept on infrastructure that will be obsolete by Q4. They're making technical commitments based on standards that don't exist yet. The CMOs presenting this as "transformation" are actually gambling with quarterly budgets.
What CMOs Actually Did vs. What They Say They Did
The BCG finding is specific. When they asked CMOs to describe their agentic AI progress, they got impressive narratives. Organizational changes. New tools. Pilot programs.
When they asked what measurable change had occurred in how campaigns are managed, the gap emerged. Most of the "progress" was repositioning existing work or adding new labels to old processes.
Some teams genuinely launched agentic pilots. Fewer actually integrated the learnings into production workflows. Almost none rebuilt their analytics to handle a world where clicks are dead and discoverability is everything.
That's not transformation. That's a PowerPoint deck masquerading as strategy.
The Real Deadline
You have roughly four quarters before agentic agents move from beta to mainstream consumer use. That's when this stops being interesting and becomes existential.
When that happens, a brand's value to an agent depends entirely on machine-readability, factual credibility, and consumer override options. All of those require rebuilding how you organize marketing data, how you structure product information, and how you think about brand trust.
Most CMOs are still thinking about campaign creative. That's fine. Just know that by the time next year's budget season hits, creative spend might not move the needle on agent discoverability.
The ones preparing for that transition aren't talking about it yet. They're quietly rebuilding their data architecture, cleaning their product feeds, and studying how agents actually evaluate brands.
The ones still talking about agentic AI transformation on earnings calls are buying time they don't have.
Want to prepare before the Q4 shift hits? Start with three things: audit your product data for machine-readability, map which brand attributes survive factual evaluation, and design for consumer override from day one.