Imagine it's 21 December. A mother's parcel, a present for her daughter, still hasn't arrived. She opens the retailer's chat. The bot tells her the order is on its way. She asks when it will arrive. The bot tells her the order is on its way. She types "speak to a person". The bot offers to help her track her order. Twenty minutes later she gives up, posts about it publicly, and the retailer loses a customer it had spent years winning. The scenario is hypothetical, but the frustrating part is the point: a human agent could have solved it in two minutes, with a free upgrade to next-day delivery.
Holiday customer service is where automation earns its keep and where it does the most damage. Contact volumes surge just as customers are most stressed, deadlines are real, and every conversation is about a gift someone cares about. Bots are brilliant at some of that work and terrible at the rest. The dividing line is simple enough to remember at 2am on a Sunday in December: automate the questions, keep the problems human. And the bot's most important job at peak isn't answering; it's knowing when to hand over.
This post covers customer service at peak. Chatbots for lead capture are covered in AI chatbots in 2026. The legal points below are general information, not legal advice.
Why peak is different
Four things change in November and December. Volume rises sharply, often several times normal levels in the busiest weeks. Deadlines matter: "where is my order" on 3 November is curiosity, on 21 December it's anxiety. Emotion runs higher, because the purchase is a gift and the stakes feel personal. And new kinds of contact appear: gift receipts, exchanges by people who didn't buy the item, delivery to different addresses, and a January wave of returns.
That combination makes automation tempting, and necessary for many businesses, but it also raises the cost of getting it wrong. A slow answer in March is an annoyance. A bot loop in December can cost a customer for good.
BabyLoveGrowth
AI platform that writes, auto-publishes SEO content and builds backlinks to grow Google & AI-search traffic.
Best for: Autopilot AI SEO content & AI-search visibility
What the 2026 data says
Customer attitudes pull in two directions, which is actually quite helpful. A December 2025 SurveyMonkey study of 2,017 US adults found that 79% strongly prefer interacting with a human over an AI agent. Yet consumers now expect instant resolutions, according to Zendesk's CX Trends 2026 report. Read together, the findings suggest people want instant answers to simple questions and humans for anything that matters.
The same report, based on responses from more than 11,000 consumers and business leaders across 22 countries, found that 74% of consumers get frustrated when they have to repeat information, which is exactly what poor bot-to-human handoffs cause. Gartner has forecast that agentic AI will autonomously resolve 80% of common customer service issues without human intervention by 2029, but it has also predicted that none of the Fortune 500 will have fully eliminated human customer service by 2028. Both are forecasts rather than findings, but read together they suggest automation is growing fast and humans aren't going away.
What to automate, and what to keep human
| Contact | Handle with | Why |
|---|---|---|
| Where is my order? | Bot, with live tracking | Clear answer from real data |
| Delivery cut-offs and options | Bot | Factual and time-critical |
| Returns policy, starting a return | Bot, exceptions to a human | Routine, but exceptions need judgement |
| Gift receipts and exchanges | Bot | Simple and high volume |
| Product questions and sizing | Bot, then human if complex | Simple ones are easy; nuanced ones aren't |
| Late gift near a deadline | Human, prioritised | Needs empathy and authority to fix |
| Damaged or wrong item | Human | Money, emotion and exceptions |
| Billing disputes, complaints, fraud | Human | Sensitive, risky, often angry |
The pattern: questions have clear answers that can be verified from data or policy, and bots handle them well. Problems involve money, emotion, exceptions or urgency, and need a person with judgement and the authority to fix things. In practice, bots tend to do best on structured tasks, such as tracking and refund status, and struggle more with complaints and billing disputes. That's a practical pattern rather than a measured benchmark, but it matches the split above.
What automation does for capacity
Getting the split right is also what makes peak manageable for your team.
A store normally handles 1,000 contacts a week. In peak week, that triples to 3,000.
Around 55% are routine questions: tracking, cut-offs, returns policy, gift receipts. That's 1,650.
A bot connected to live data resolves 70% of those: about 1,155 contacts.
→ Contacts needing a human fall from 3,000 to 1,845, about 1.85× normal volume instead of 3×
→ If every contact took the same 8-minute average (a simplification), that's roughly 400 hours down to about 246
The catch: the contacts left for humans are the hard ones, so they're likely to take longer than 8 minutes and real hours will be higher than 246. Treat this as an illustration, not a staffing forecast, and plan for complexity, not just count.
Your mix will differ, so tag a sample of last year's holiday contacts before setting targets.
The principle Automate the questions, keep the problems human. At peak, a bot's most important job isn't answering; it's knowing when to hand over.
Design the handoff before anything else
The biggest failure in holiday support isn't bots that can't answer; it's bots that won't let go. Build clear handoff triggers:
- The customer asks for a person, in any wording.
- The same question comes up twice, or the bot's answer didn't resolve it.
- Frustration or urgency appears in the language.
- Money is involved: refunds, charges, compensation.
- A deadline is close: a delivery problem within a few days of an important date.
- The customer is valuable or vulnerable: a high-value order, a loyal customer, or anyone who needs extra support.
Then make the handoff good. Pass the full conversation and order details to the agent, so the customer never repeats themselves. Be honest about wait times, and offer a callback or email if queues are long. Never hide the route to a person or trap customers in loops. Here's what a good handoff might sound like for the hypothetical mother in the opening: "I'm sorry your gift hasn't arrived, and I know the date matters. I'm passing you to Sam in our team now, with your order details, so you won't need to repeat anything. The current wait is about four minutes, or I can arrange a call back within the hour. Which would you prefer?" It's honest, it acknowledges the stakes, and it hands over with context. The same principles of respectful conversation design apply as in designing chatbot flows that don't frustrate users.
Keep the bot honest
A bot that makes things up is worse than no bot. Connect it to live order and tracking data, and to an accurate knowledge base with this season's cut-offs, holiday returns rules and opening hours. Tell it to say "I don't know, let me connect you" rather than guess. The risk is real: in Moffatt v. Air Canada (2024), British Columbia's Civil Resolution Tribunal found Air Canada liable for negligent misrepresentation after its website chatbot wrongly told a customer they could claim a bereavement fare retroactively. It's a small-claims decision on its own facts, not a universal rule, but it rejected the idea that the airline wasn't responsible for its chatbot's information. The same discipline of keeping AI accurate about your brand is covered in keeping AI accurate about your brand.
Tell people they're talking to a bot. Article 50 of the EU AI Act, which applies from 2 August 2026, requires providers of AI systems that interact directly with people to design them so people are informed they're dealing with AI, unless that's obvious. Businesses deploying a vendor's bot have different duties, so check your provider's compliance and make the disclosure clear in the chat itself. California's narrower bot law targets bots used online to mislead people in California about their artificial identity in order to drive a sale or influence a vote; clear disclosure avoids liability. Beyond compliance, honesty builds trust. Your wider rules on AI use and disclosure belong in the policy set out in AI governance for marketing teams. And be realistic about what newer, more autonomous AI agents can safely do on their own, a question explored in what agentic automation can and cannot do yet.
Prevent contacts before they happen
The cheapest support contact is the one that never happens. Many "where is my order" questions disappear when customers get clear, proactive updates: shipping confirmation, delay alerts and delivery-day notices by email or text, following the timing and consent rules in SMS compliance basics. Clear delivery promises on product pages prevent another wave, as covered in shipping and delivery promises, and well-built post-purchase flows answer questions before customers think to ask. Publish holiday returns rules plainly, including how partial returns of gift sets work, as covered in holiday bundles and gift sets, since returns questions spike in January, and the choices involved are covered in how retailers are rewriting returns policy.
Get ready in the next two weeks
With Black Friday on 27 November, the preparation window is short. Review last year's holiday contacts and tag the top twenty reasons. Map how each should be handled, following the approach in mapping workflows before you automate them. Update the knowledge base with this season's dates and policies. Test the bot on real questions, including awkward ones, and check every handoff works. Staff the human queue for the hard cases and brief agents on what they can authorise, such as shipping upgrades or goodwill gestures. And set a daily check during peak to review failed conversations and fix gaps quickly. Teams that want automation and human support designed together often work with a specialist marketing automation team.
Measure resolution, not deflection
Many teams judge bots on how many conversations never reach a human. That number is easy to inflate by making escalation difficult, which is exactly how customers end up trapped. Measure instead whether problems were actually solved: resolution rate, repeat contacts within a few days, satisfaction by contact type, how quickly escalations are picked up, and complaints that mention the bot. A bot that hands over quickly and well is doing its job.
The short version
Holiday customer service combines higher volumes, real deadlines, stronger emotions and new kinds of contact, which makes automation both necessary and risky. Customers want instant answers to simple questions and people for anything that matters: a December 2025 SurveyMonkey study found 79% of US adults strongly prefer a human over an AI agent, and Zendesk's CX Trends 2026 found 74% of consumers get frustrated when they have to repeat information. Automate the questions, keep the problems human. Let bots handle order tracking, delivery cut-offs, returns policy, gift receipts and simple product questions using live data; route late gifts near deadlines, damaged items, billing disputes, complaints, fraud and vulnerable customers to people. Design handoffs first, with clear triggers and full context passed along. Keep the bot honest, since a 2024 tribunal decision held Air Canada liable for its chatbot's wrong answer, and make AI use clear; EU AI Act transparency rules for providers apply from 2 August 2026. Prevent contacts with proactive updates, prepare in the next two weeks, and measure resolution rather than deflection.
Is your holiday support ready for peak week?
We design chatbot and human support flows that resolve problems without trapping customers.
Explore Marketing Automation →Frequently asked questions
What should a customer service chatbot handle during the holidays?
Questions with clear, verifiable answers. During the holidays, a large share of contacts are routine: where is my order, what are the delivery cut-off dates, what is the returns policy, how do I start a return or exchange, can I get a gift receipt, and what are the opening hours. A chatbot connected to live order data and an accurate, up-to-date knowledge base can resolve many of these instantly, at any hour, which frees human agents for harder cases. Bots can also help with simple product questions, such as sizes or compatibility, as long as they can hand over when questions become complex. The key requirement is that the bot answers only from verified information, such as real tracking data and current policies, rather than guessing, and that it can pass the customer to a person quickly when needed.
What should stay human in holiday customer service?
Problems that involve emotion, money, exceptions or urgency. During the holidays, that includes a gift that has not arrived close to a deadline, a damaged or wrong item, billing disputes or double charges, complaints and angry customers, account security or suspected fraud, and anyone who needs extra support because of disability, age or circumstances. These situations usually need judgement, empathy and the authority to make exceptions, such as upgrading shipping, sending a replacement or issuing a refund. Surveys also suggest many customers prefer to speak with a human for complex or sensitive issues. A good holiday setup lets bots handle routine questions and routes these harder cases quickly to trained people, with the full context passed along so the customer does not have to repeat themselves.
When should a chatbot hand over to a human agent?
Whenever the customer asks, and whenever signals suggest the bot cannot resolve the issue well. Clear triggers include a direct request for a person, the same question being asked twice, frustrated or angry language, a problem involving money such as refunds or charges, a delivery issue close to an important date, a high-value order or a loyal customer, and any topic the bot is not designed to handle, such as complaints, security or legal questions. The handoff should pass the full conversation and relevant order details to the agent, tell the customer honestly how long they may wait, and offer alternatives such as a callback or email if queues are long. Never trap customers in loops or hide the option to reach a person.
Do businesses have to tell customers they are talking to a chatbot?
In some places and situations, yes, and it is good practice everywhere. Under Article 50 of the EU AI Act, which applies from 2 August 2026, providers of AI systems that interact directly with people must design them so that people are informed they are interacting with AI, unless it is obvious from the context. Businesses deploying a vendor's chatbot have different duties, so they should check their provider's compliance and make the disclosure clear in the customer experience. California's narrower bot law targets bots used online to mislead people in California about their artificial identity in order to drive a sale or influence a vote; clear disclosure avoids liability. Separately, in a 2024 small-claims decision, British Columbia's Civil Resolution Tribunal found Air Canada liable for negligent misrepresentation after its website chatbot gave a customer incorrect information about bereavement fares. Clearly labelling the bot, making it easy to reach a human, and making sure the bot answers only from accurate information all reduce risk. This is general information, not legal advice.
