mar-techFeatured

Three converging data points are reshaping how senior marketers should think about AI's role in the purchase funnel

Tuesday, April 28, 2026·5 min read·10 sources

The Signal

Three converging data points are reshaping how senior marketers should think about AI's role in the purchase funnel. A survey of 1,009 U.S. consumers by Exploding Topics found that 77.6% have used AI to assist with shopping decisions in the past six months, with 43.21% doing so weekly or more. ChatGPT leads as the tool of choice among AI shoppers (77.56%), followed by Google Gemini (58.21%). AI has directly influenced 68.64% of users to make a purchase they otherwise would not have made. The trust gap, however, is sharp and consequential. The modal amount consumers would authorize an AI agent to spend autonomously is $0. Thirty-one percent would authorize no autonomous spend at all, and 51.45% are uncomfortable with AI tools storing card details. "Skeptical" and "suspicious" are the two most common attitudes toward agentic checkout tools such as ChatGPT's Instant Checkout — a product OpenAI has already walked back. Simultaneously, a Semrush analysis of billions of visits across 50,000-plus sites confirms that traffic channel mix is shifting materially as AI intermediates the discovery phase. MIT Technology Review's reporting with Databricks and Infosys underscores the infrastructure gap: enterprise AI effectiveness remains constrained by fragmented, ungoverned data estates, with 95% of new AI projects failing to generate measurable business value.

Why It Matters

The trust gap in agentic commerce isn't a consumer education problem—it's a product-market fit problem that the industry keeps misdiagnosing. OpenAI didn't walk back Instant Checkout because users didn't understand it; they walked back because users understood it perfectly and said no. Treating skepticism as a literacy deficit that marketing can fix is the wrong frame entirely. What surface-level coverage misses is the structural reason trust is so fragile here: consumers already suspect AI commerce tools are designed to serve brands and AI companies first. Only 14% believe they're the primary beneficiary. That's not ambivalence—that's a coherent worldview. And it's correct. The commercial architecture of these tools, from sponsored placements to monetized recommendations, confirms the suspicion. Asking consumers to hand over autonomous spend authority to a system they believe is working against their interests isn't a trust problem you solve with better UX or clearer disclaimers. The deeper pattern is that AI is successfully disintermediating the discovery phase while failing to capture the transaction layer—and these two dynamics are moving in opposite directions. Semrush's traffic data confirms the discovery shift is real and accelerating. But the $0 modal spend authorization tells you the transaction layer remains human-guarded, likely for years. Marketers who treat this as a temporary gap to be closed are misreading the signal; the more productive read is that AI has permanently restructured where brand influence gets won, and it's earlier in the funnel than most organizations are currently optimizing for. For decision-makers, the strategic implication is unambiguous: GEO investment and data infrastructure that feeds AI recommendation layers aren't optional bets on the future—they're the primary battleground right now.

The Exploit

🎯

Opportunity

Brand and demand-gen leaders at mid-to-large B2C companies with clean, structured product data estates are positioned to capture disproportionate share of AI-mediated discovery right now. With 77.6% of consumers already using AI for purchase decisions, the recommendation layer is live and being won—by whoever feeds it best. The 6-month window before AI discovery optimization becomes table stakes is still open.

⚠️

Risk

Teams misallocating Q3 and Q4 budget toward agentic checkout optimization—better UX, autonomous spend incentives, stored-payment conversion—are burning resources on a layer consumers have explicitly locked. The failure mode is measurable: declining organic visibility in AI-generated recommendations while competitors who invested upstream capture the consideration set before the human-guarded transaction even begins.

🚀

The Move

Commission a GEO readiness audit against your top 20 revenue-driving product categories before Q3 2026 budget lock—specifically mapping where your structured data, schema markup, and content architecture either surface in or disappear from ChatGPT and Gemini recommendation outputs. The CMO owns this decision; success criteria is a clear delta score showing which categories are invisible to AI discovery and a funded remediation plan entering Q3 planning.

First-Mover Advantage

Gains

Risks

Window

Winners & Losers

Winners↑

↑

Generative Engine Optimisation (GEO) Agencies and Consultancies

The confirmation that 44% of consumers use AI as their starting point in the purchase journey — before ever touching a retail site — makes GEO the primary battleground for brand influence, not a speculative future investment. The mechanism is direct: brands that don't appear organically in ChatGPT and Gemini responses are invisible at the moment of highest intent, and agencies that can deliver that visibility are now structurally indispensable. These firms should aggressively productise GEO audits, AI citation monitoring, and structured data optimisation as core retainer services rather than add-ons.

↑

Enterprise Data Infrastructure and Governance Vendors

The MIT/Databricks/Infosys reporting crystallises a durable competitive moat: proprietary, well-governed data is the differentiating input that frontier AI models cannot replicate, and 95% of enterprise AI projects failing without it validates the infrastructure gap as a primary sales conversation. Vendors like Databricks that unify analytical and operational data estates are positioned at the centre of every AI roadmap that actually ships to production. The strategic response is to lead with governance and value-measurement frameworks — the 3M methodology framing resonates precisely because it converts AI ambiguity into board-level accountability.

↑

Upper-Funnel and Discovery-Layer Advertising Platforms

As AI intermediates the discovery phase and compresses traditional middle-funnel touchpoints, platforms that operate where brand consideration is formed — contextual placements within AI interfaces, sponsored positions in AI Overviews, and recommendation-layer inventory — gain structural pricing power. The Semrush traffic channel data confirming the discovery shift, combined with the consumer survey showing AI directly influenced 68.64% of users to make a purchase, means upper-funnel brand presence now has a more direct conversion mechanism than historically modelled. These platforms should invest in attribution tooling that connects AI-layer impression data to downstream purchase outcomes to capture the budget currently sitting in confused performance marketing teams.

↑

First-Party Data Activation and Clean Room Technology Providers

With 51.45% of consumers uncomfortable storing card details with AI tools and only 14% believing AI commerce serves their interests, the trust deficit creates a structural premium on brand-owned data relationships and privacy-preserving data collaboration. Clean room and first-party activation vendors are positioned to capture budget as brands realise that feeding proprietary customer data into AI recommendation layers — without surrendering it to platform intermediaries — is the mechanism by which they retain influence over AI-mediated discovery. The pitch sharpens considerably when framed as feeding the GEO layer with governed, high-quality signals rather than generic AI experimentation.

Losers↓

↓

Agentic Checkout and Autonomous Transaction Tool Builders

The $0 modal autonomous spend authorisation is not a temporary adoption gap — it reflects a coherent consumer belief that agentic commerce tools are designed to serve platforms and brands ahead of shoppers, a view held by 86% of respondents including the tools' most frequent users. OpenAI's walkback of Instant Checkout is the leading indicator: the product-market fit problem is structural, not solvable through UX iteration or consumer education campaigns. Builders in this category should pivot toward discovery-and-handoff architectures — tools that surface options and route to human-controlled checkout — rather than continuing to push autonomous transaction capabilities against an audience that has explicitly declined them.

↓

Bottom-Funnel Performance Marketing Automation Teams

Teams optimised for capturing conversion intent at the bottom of the funnel face a dual squeeze: AI is restructuring where brand influence is won (earlier, in the discovery layer) while simultaneously failing to capture the transaction layer, meaning the conversion moment remains human-guarded but the preceding decision is increasingly made elsewhere. Budget and attention allocated to retargeting, dynamic product ads, and last-click optimisation is increasingly chasing a consumer whose consideration set was already closed before they reached the brand's owned channels. The defensive move is to reallocate testing budgets toward measuring AI citation frequency and brand sentiment within AI outputs, and to build internal capability for GEO before procurement cycles force the conversation.

Strategic Outlook

The timing here is not accidental. OpenAI's walkback of Instant Checkout follows a recognizable pattern: aggressive agentic commerce feature launches meeting consumer resistance, followed by quiet retreats to safer discovery-layer positioning. Google executed a similar recalibration with its Shopping Graph integrations, and Amazon has spent years discovering that consumers will tolerate algorithmic recommendations but bristle at the moment autonomous systems gain financial access. The commerce transaction layer has consistently been where consumer trust collapses, regardless of how much goodwill the discovery layer accumulates. What's different now is the data infrastructure dimension arriving simultaneously. The MIT/Databricks/Infosys analysis puts a specific number on enterprise AI failure rates — 95% of new projects generating no measurable business value — which reframes the GEO conversation entirely. Marketers competing for AI recommendation visibility are essentially competing for placement in systems that most enterprises cannot yet feed reliably. The brands winning in AI-mediated discovery right now are disproportionately those that solved data governance problems two or three years ago, not those launching GEO initiatives today. The Semrush traffic channel analysis adds the third dimension: the shift away from direct and organic search is accelerating precisely as the transaction layer calcifies. That combination — eroding traditional discovery channels, frozen agentic transaction layer, infrastructure gap persisting through at least Q3 2026 — means the window for influence has compressed into a narrower, earlier funnel moment than most marketing organizations are currently staffed or structured to own.