Retail in the Breached Quadrant

Senior marketing leaders across industries see the wave coming. In BCG x Moloco’s survey of 283 executives across 15 verticals conducted in October 2025, 67% expect high AI-driven disruption to their consumer journeys. Retail leaders see something different: only 36% share that concern.

The gap runs deeper than general sentiment.

Retail isn’t just calm about AI disruption. It’s the calmest sector in the study.
‍
BCG’s X Moloco Consumer AI Disruption Index tells a different story. The index scores verticals on two dimensions: the extent to which AI can disintermediate the customer relationship, and the strength of that relationship today. Retail lands in the “breached” quadrant alongside travel and news. A shopper can now ask an agent for a recommendation, evaluate options, and complete the purchase without reaching a retailer's site.

Retail has real assets: fulfillment networks, loyalty programs, first-party data. But assets don’t change the exposure score. Either retail leaders see something the index missed, or the disruption will arrive before the urgency does.

The Gap is Already Visible

Protocols from Google and OpenAI are standardizing agent-to-merchant transactions. The infrastructure that lets an AI agent discover inventory, verify claims, and complete a purchase is live, and 30% of consumers now say they're comfortable letting AI buy on their behalf.

The experience gap tells you where this lands. A shopper asks an agent for a standing desk that fits a small apartment with good cable management. The agent walks them through weight capacity, height range, assembly complexity. They settle on a recommendation. Then they click through to the product page: a static spec sheet, reviews sorted by recency, and a chat widget in the corner that rarely gets used. The conversation they just had was intelligent. The experience they landed on asks them to start over.

One retail CMO described the shift:

"That initial click [to our website] is not happening because [consumers are] getting the information they need directly [from AI overviews]…

I think the top of funnel is gonna be the highest disruption... but honestly it's full funnel disruption, because even at the bottom of funnel… that journey is being disrupted."

The risk isn't that shoppers stop visiting retail sites. It's that they arrive having already made the decision, with the retailer reduced to fulfillment.

The Asset They are not Using

If service disruption is the real exposure, the natural question is whether retailers have any structural advantage. They do but most are not using it.

ChatGPT serves every vertical, every shopper, every query. Breadth is the product. That’s a strength and a constraint: the economics of a generalist model don’t justify going deep on any single category when you’re optimizing across all of them.

Retailers operate under different math. You hold the post-purchase truth: the returns, the support tickets, the fulfillment exceptions. You know that return rates spike when two specific products are bought together, a bundle incompatibility no review mentions, because you see the pattern across thousands of transactions. You know a “universal” smart home hub has connectivity issues with Ring Gen 2 doorbells because you see the support tickets. You know which mattress actually works for side sleepers with back pain because you’ve tracked which ones come back.

This is one of the drivers for why the index identifies retail as “defensible.” The first party outcome data that reveals what actually works versus what the spec sheet says should work.
‍
But defensible is not defended. One CMO admitted the gap directly: “We don’t even have a customer data platform right now. Our customer data is split up among a bunch of different things... without that, we can’t really use an AI-based re-engagement tool.”
‍
Encoding looks different by category. A home goods retailer turns return patterns into "these two products don't work well together" surfaced before the shopper adds both to cart. An electronics seller turns support tickets into “known compatibility issue” before the shopper makes the wrong choice. A furniture site turns behavioral sequences into guided entry points that cut 200 results to 4.

Each of these stops offering inventory and starts offering intelligence. The shopper comes back because the site understood what they needed, not just what they typed. That’s how retail moves out of the breached quadrant.

of retail marketing leaders say they’re prioritizing first-party data capture

expect high disruption

see serious service takeover risk

Intent without urgency is a recipe for pilots that stall, roadmaps that slip, learning cycles that never compound.

We expect retailers that are experimenting often will discover which catalog attributes predict conversion, which interventions earn trust, which signals to protect. Whereas retailers that rarely experiment may potentially never learn those lessons. By the time agent-referred traffic reaches a substantial portion of sessions, the defaults will be set: which retailers agents recommend, which catalogs they trust, which data they rely on.

The defaults are forming now.

Over the course of October 2025, BCG x Moloco’s Consumer AI Disruption Index assessed 17 consumer-facing verticals through a quantitative and qualitative study (note that two verticals only participated in the qualitative and not the quantitative portion of the study) of 238 senior marketing leaders across five regions, combined with app performance data from more than 3,200 apps. The full Moloco Readiness Guide includes a diagnostic for retailers seeking to assess their catalog structure, technical infrastructure, and organizational coordination. Full methodology available in ‘The AI Disruption Index: How AI is Reshaping Consumer Discovery’.