transformationplaybook.aiSqwyz
AI Wire · 2 NOV 2026

McKinsey: 17% of US AI Users Have Bought Through an AI Tool, Up From 10%

Close-up of rows of server cooling fans in a data centre, lit in pink and blue

Photo: Winston Chen on Unsplash

26 Oct to 2 Nov 2026
17%

of US AI users have bought directly inside an AI tool, up from 10% a year earlier, McKinsey reports.

McKinsey's June 2026 consumer survey covers about 21,000 people in 11 markets. Globally 14% of AI users have bought through an AI tool. Retailers' own AI assistants trail stand-alone tools at discovery but are marginally ahead at checkout (15% against 14%, though the two questions were worded differently). That is likely because retailers already hold live stock, payment and delivery data.

The week in three lines

  1. McKinsey finds 17% of US AI users have bought directly in an AI tool, up from 10% a year earlier. Retailer assistants are marginally ahead of stand-alone tools at checkout (15% against 14%, though the questions were worded differently), likely because retailers already hold live stock, payment and delivery data.
  2. Deloitte and McKinsey describe the same gap. Of Deloitte's respondents, 81% report faster decision-making and execution but only 25% are deeply transforming the business, and in McKinsey's procurement survey only 2% have AI embedded at scale.
  3. McKinsey offers a simple test of agent readiness: could a retailer's systems accurately state what a product costs, whether it is in stock, when it could arrive and whether the agent can buy it now?

Enterprise AI: speed versus redesign (Deloitte, 3,500 leaders)

Report faster decision-making and execution81%
Redesigned jobs around AI, at least moderately60%
Report revenue gains from AI28%
Using AI to deeply transform the business25%
Report a significant shift in decision-making authority15%

Shopping with AI (McKinsey consumer survey, June 2026)

Consumers who used a stand-alone AI chat or search tool56%
Consumers who used a retailer's AI assistant27%
US AI users who bought directly in an AI tool17%
Global AI users who bought directly in an AI tool14%

The papers

McKinsey28 Oct 2026Supply chain

McKinsey: over half of consumers now use AI to shop, and retailers' own assistants catch up at checkout

A June 2026 survey of about 21,000 consumers in 11 markets finds 53% used AI for shopping in the past three months, and 17% of US AI users have bought directly in an AI tool, up from 10%. Retailer assistants trail stand-alone tools by up to two times earlier in the journey but are marginally ahead at checkout (15% against 14%, though the two questions were worded differently).

Why it matters

Retailer assistants likely hold their ground at checkout because retailers already hold live stock, payment and delivery data, so availability and promise-date accuracy are likely to become conversion inputs. Grocery and beauty, where online penetration is low, are already buying through AI. Substitution, slot capacity and returns handling need testing against agent-led demand.

Read the source
The playbookFree toolkit
McKinsey28 Oct 2026Supply chain

McKinsey: agentic AI in sourcing pays when aimed at sourcing decisions, and requisitions stay with simple automation

McKinsey cites a survey of procurement leaders in which 80% plan to invest more in AI but 2% have it embedded at scale, and says agents aimed at sourcing decisions can add 3% to 6% savings on external spend. The examples are anonymous and the value comparison is McKinsey's own calculation.

Why it matters

Cost per decision against the value at stake decides where agents belong, so routine requisition traffic stays on simple automation. McKinsey says contract intelligence is the stage most often skipped, and negotiations depend on it. It advises replaying 12 months of past events and running in shadow before any live decision. The 3% to 6% savings are McKinsey's own case from anonymous examples and have not been independently confirmed.

Read the source
The playbookFree toolkit
Deloitte AI Institute29 Oct 2026Value gap

Deloitte: AI makes work faster, but only 15% report a significant shift in who decides

A survey of 3,500 director to C-suite leaders in 25 countries (May to June 2026) finds 81% report faster decisions, 60% have redesigned jobs to at least a moderate degree and 15% report a significant shift in decision-making authority. 40% let AI act without human approval at least moderately, and 23% say accountability is unclear when it goes wrong.

Why it matters

If approval limits and escalation paths are unchanged, faster analysis just queues up. Accountability for agent actions is a live question, with 23% saying it is unclear when AI produces a negative outcome. Only 44% track per-unit inference cost.

Read the source
The playbookFree toolkit
BCG30 Oct 2026Chatbots

BCG: CPG brands face AI agents across the whole funnel, and most have no capability yet

BCG says 41% of CPG companies have begun building agentic capabilities and 59% report none, and that 87% of retailers do not share customer data with agents at all. Its revenue gains from pilots are BCG's own figures, with no sample sizes shown.

Why it matters

Where retailers withhold customer data from agents, brands are likely to have to negotiate agent access and promotion data feeds as part of the joint business plan. Real-time availability and promotion data become a bargaining asset, and building them is as much a supply chain job as a marketing one.

Read the source
The playbookFree toolkit
Forrester30 Oct 2026Chatbots

Forrester: customers' AI agents won't give up, so hidden service demand will surface

An analyst argues that personal AI agents will chase problems customers used to abandon, so contact volumes may rise and containment metrics will mislead. The post is opinion with one adoption figure and no original survey data.

Why it matters

Late deliveries, refunds and missing items that customers now abandon are likely to resurface as contacts. A service automation case built on today's volumes should be tested against the abandoned volume, with delivery and returns teams owning the fix at source.

Read the source
The playbookFree toolkit

Also published

What nobody is saying

BCG and McKinsey put the AI shopper at very different levels, and the figures don't measure the same thing. BCG's 4% is the share of US consumers who say they would let an agent complete a purchase alone (December 2025). McKinsey's 14% globally, and 17% in the US, is the share of AI users who report having completed a purchase directly inside an AI tool (June 2026). One measures stated willingness among all consumers, the other reported behaviour among AI users. McKinsey also relays two platform claims without method: Amazon says Alexa for Shopping users are 60% more likely to buy, and Shopify says AI searches using its structured catalogue convert at roughly twice the rate of searches on scraped or stale data. Both come from companies selling the tools.