82% of Shoppers Want Brands to Fix Broken AI Promises

A new holiday-season report from Forethought finds that 82% of shoppers expect brands to make things right when AI-powered customer service falls short of what was promised. The finding lands as retailers lean more heavily on AI chatbots, virtual agents and automated support tools to handle the surge in holiday shopping inquiries.

The report signals a widening gap between what AI-driven service tools promise shoppers — fast answers, accurate order tracking, seamless resolutions — and what many of those tools actually deliver once volume spikes. Rather than excusing AI shortcomings as growing pains, the overwhelming majority of consumers surveyed say accountability still sits squarely with the brand.

This is a trust story as much as a technology story. As retailers deploy AI across more of the customer journey — pre-purchase questions, order tracking, returns, complaint handling — they are effectively making promises on the technology's behalf. Forethought's finding suggests customers don't distinguish between a human agent's mistake and a bot's mistake; they simply expect the outcome they were promised, and they expect the brand to fix it when that promise breaks.

For experience and digital transformation leaders, this reframes AI rollout as a service-design problem, not just an engineering one. Deploying AI without a reliable human escalation path, clear expectation-setting, and a fast remediation process invites exactly the kind of broken-promise moments shoppers say they won't quietly accept — especially during the highest-stakes, highest-volume weeks of the year.

— 82% of shoppers surveyed expect brands to make good when AI-powered service promises go unmet, according to Forethought's holiday report.

Retailers have spent the past two years selling AI as a reliability upgrade — faster answers, fewer errors, less friction. This data point is a reminder that customers never actually bought "AI"; they bought a promise, and they will hold the brand to it regardless of what's behind the curtain.

Most companies treat AI failures as a technical footnote — a bug to patch quietly before the next release. Customers don't grant that grace period, and holiday volume makes every gap visible fast. The fix isn't more disclaimers about AI being "in beta"; it's designing the recovery moment as carefully as the AI moment itself — clear human escalation, fast remediation, and zero ambiguity about who owns the fix when the bot gets it wrong. Brands that treat AI accountability as someone else's problem will find their customers disagree, loudly, in the exact week they can least afford it.