Holiday Shopping 2025: 78% Used AI, Yet Only 29% Saw Better Service-Humans Still Lead

AI Took Over Holiday Customer Service in 2025-But Humans Still Win Where It Counts

AI was everywhere this holiday season, but it didn't fully win over shoppers. According to the Liveops 2025 Holiday AI & Customer Service Report, 78% of consumers used AI or automation for support, and 73% leaned on it more than last year. Still, only 29% say it improved their experience-and 54% say humans delivered better service.

If you lead a support team, the message is clear: speed is up, trust is not. The opportunity isn't "more AI"; it's better orchestration between automation and people.

What the numbers say

• Usage: 78% interacted with AI; 73% used it more than in 2024.
• Channels with the biggest automation lift: online chat and website help (61%), phone lines (39%), automated emails/texts (36%).
• Speed: 85% said AI made service faster and more accessible.
• Experience gap: 29% reported a better holiday experience, 14% worse.
• Human edge: 54% say human agents still provide better service.
• Breakdowns: 55% escalated an AI-handled issue to a human; 45% said AI didn't understand the problem.
• Transparency: Only 22% said companies clearly disclosed AI use; 69% want brands to disclose it.
• Generations: 89% of Gen Z used AI vs. 60% of Boomers.
• Future use: Just 17% want more AI next year; 32% want less.

Why this matters for support teams

• AI wins on speed but loses when context, empathy, or judgment are needed. That's why escalation rates are high.
• Trust breaks when customers can't tell if they're talking to a bot-or when they can't reach a human fast.
• Channel strategy matters. Automation is working in chat and web help, but phone and SMS need smarter guardrails.
• Age and preference drive channel expectations. Younger customers tolerate automation; older customers expect easy access to a person.

Action plan for your next peak season

• Design for hybrid from the start: route simple, high-volume intents to AI; hand off complex or emotional cases to humans with full context.
• Make escalation effortless: visible "talk to a person" options, skill-based routing, and warm transfers with conversation history.
• Set clear transparency norms: label AI interactions and explain why it's used. This reduces confusion and rebuilds trust.
• Measure what matters: pair speed metrics (AHT, time-to-first-response) with outcome metrics (FCR, CES, NPS, recontact rate, containment without recontact).
• Harden failure modes: if confidence is low or the customer restates twice, escalate. No dead ends, no loops.
• Tune continuously: analyze reason codes for escalations, update intents, improve prompts, and retrain workflows weekly during peak.
• Coach for empathy at the handoff: agents should acknowledge friction with automation and resolve fast. Recovery beats perfection.
• Segment by customer profile: default to chat/self-serve for digital-savvy customers; keep phone paths short for Boomers and high-value accounts.
• Protect privacy: redact PII in transcripts, restrict model access to the minimum, and log every automated decision.

Operational patterns that work

• First-contact triage: intent detection + verification + quick self-serve paths with a human failsafe.
• Knowledge that adapts: AI answers grounded in your approved content, with freshness checks during promotions and policy changes.
• Unified context: one timeline across bot and human, including order data, recent tickets, and channel history.
• Clear ownership: who owns what when AI is wrong? Set SLAs for bot-caused escalations to protect CSAT.

Compliance and trust

• Publish an AI use statement and stick to it. Disclose when AI is involved and how to reach a human.
• Adopt risk controls aligned with frameworks like the NIST AI Risk Management Framework.

Team enablement

• Train agents on AI-assisted workflows, not just tools: reading bot transcripts, validating AI-suggested actions, and closing the loop.
• Level up prompts and knowledge design so automation answers reflect your brand tone and policies. If you need structured upskilling, explore AI courses by job role.

KPIs to watch weekly in peak

• AI containment vs. recontact within 7 days
• Escalation reasons and time-to-human
• Resolution rate gap: AI vs. human for the same intents
• CSAT/effort score delta pre- and post-escalation
• Refund/return impact where AI handled preauthorization

Bottom line

Shoppers aren't asking for more automation. They want AI that understands their issue, resolves it, and gets them to a person without friction when needed.

Build for that reality. Blend automation with human expertise, be transparent, and optimize for resolution-not just speed. For the source data, see the Liveops 2025 Holiday AI & Customer Service Report at liveops.com.