How to use omnichannel AI to improve the customer experience

Your support team probably answers the same handful of questions all day about where an order is, whether a customer can return something, and if you have a shirt in another size.

The questions are easy, but they pile up. While your team works through them, the harder conversations that actually build loyalty wait in line behind them.

The easy questions are what pushed customer service towards AI in the first place. As early as the 1990s, engineers were building basic bots that could follow simple decision trees to field them.

Those early bots were the start of what we now call AI in customer service: the use of conversational agents and automation to answer customer questions, resolve common requests, route complex issues to human agents, and handle repetitive tasks like order tracking, returns, and FAQs automatically so support teams can focus on higher-value conversations.

Today's AI customer agents are far better at that work than those early decision trees. The best ones are trained on your help doc, product catalogue, and historical and real-time customer data to resolve questions on the spot instead of pushing the customer down a menu, which is why so much routine support volume is now within reach of automation.

AI could handle up to 60% of addressable care volume and free your people for the interactions that actually need a human, according to McKinsey. Here are 7 concrete examples of how that might work in practice.

1. AI is tested and governed before it ever talks to a customer

Your AI customer service programme is much more likely to fall short if no one pressure-tests it first. Often, nobody pressure-tests it because there’s no way to try an AI agent against realistic scenarios before it's live, which unfortunately means the first stress test ends up being an actual customer.

Trust in AI comes from testing it, not hoping it behaves. AI-forward brands treat new AI the way you'd treat new code by validating it before it ships: by running simulations to test and validate how the AI agent behaves in conversations before going live, all backed by guardrails and full audit trails.

Before your AI customer agent goes live, run it through the hard cases: a frustrated loyalty-tier customer, a wrong-size exchange, a tight return-window deadline. Assess how it handles each exchange, and fix what breaks before you expose it to real people.

2. AI resolves order status questions on its own

There's one question that strikes a unique combination of fear and dread in the heart of any customer service pro: 'Where is my order?'

Known as ‘WISMO’ in the industry, these requests arrive in high volume, follow a predictable pattern, and have a clear answer sitting in your order and shipping data. The combination of an endless pile of tickets and a large amount of information for AI to refer to is why WISMO enquiries are the first thing most teams automate.

Like WISMO, a lot of support tickets arise when a customer simply can’t find something they’re already allowed to see, like their loyalty balance, a subscription renewal date, or your return and exchange policy. They want that detail the moment they think to look, usually mid-browse, and if they can’t find it immediately, they might reach out to your support team instead of continuing to search.

When someone asks where their package is or when their loyalty points expire, an AI agent with pre-built order tracking skills pulls that information directly from their customer profile, answers instantly, and closes the conversation, all without a human touching the ticket.

This kind of thing is particularly valuable during seasons like Black Friday Cyber Monday, when order enquiries surge. After enabling an AI customer agent on their website, home fragrance brand Happy Wax saw ‘a dramatic reduction in support tickets’, says Rachel Fagan, VP of marketing: over half of conversations in a 90-day period were fully resolved without any human involvement.

‘Customers get instant answers, and our team gains bandwidth for high-touch moments’, Fagan explains. ‘That’s setting us up for success this BFCM.’

3. AI handles returns, exchanges and warranty claims

Returns matter to customers, and they're a slog for your team. Each one is repetitive but hands-on, since resolving it means taking an action, not just looking something up, and that combination ties up a human for several minutes at a time.

It's also exactly the kind of work a more capable AI agent can take off their plate.

When a customer starts a return or warranty claim, an AI agent with pre-built return skills can verify their identity, check eligibility, and issue the replacement or refund on its own, escalating only the edge cases that genuinely need judgement.

Automating returns gives you back the staff hours each return or warranty claim eats up, and it gives customers the fast, painless resolution that decides whether they shop with you again. That makes for a much better post-purchase experience for everyone involved.

NANUK, a brand that makes waterproof cases, understood this firsthand. Their human customer service team was staffed with reps who’d been with the brand for over a decade, and they were wasting too much of their time on simple questions. It wasn’t the best use of their time, and it slowed down their ability to manage the kind of hands-on, long-term customer relationships that really move the needle on retention.

When NANUK launched an AI customer agent, it resolved 100% of return and exchange queries in 90 days. ‘The AI agent does not replace human agents’, says Sophie Morin, director of digital marketing and ecommerce. ‘It’s another layer of service that gives us a faster, more accurate way to answer simple questions. It allows us to bring our customer service to another level’.

4. AI recommends products and recovers revenue

Customer service AI doesn't have to be a cost centre. With full customer context, the same AI agent that answers a question can recommend products, cross-sell, and recover a sale inside a live conversation.

Picture a customer who opens an AI web chat after abandoning a shopping cart. The AI agent answers their question, spots the open cart, and sends a recovery offer in the same interaction. Or a returning shopper asks about a product, and the AI agent uses their purchase and browsing history to recommend relevant items.

When luxury apparel brand Naked Wardrobe wanted to create an elevated concierge experience on their website, they launched an AI customer agent to give their customers access to a boutique shopping experience online. In 90 days, the AI agent resolved 94% of product recommendation queries.

‘Now, AI is styling our customer, answering her questions, and selling her the right piece, all at two in the morning when our team is asleep’, says James Thorngren, director of marketing at Naked Wardrobe.

"Customer Agent could have a conversation with every customer who lands on our site, and exactly understand their needs, and make a helpful recommendation," agrees Sophie Morin, director of digital marketing and ecommerce at NANUK. 'This is not a chatbot from 2010'.

5. AI delivers always-on support across languages and channels

Support has always had two ceilings: the clock and the language barrier. A customer in Lyon, France who needs help at midnight local time has historically been out of luck if your team speaks English and works American business hours.

After reaching out about a negative experience, 81% of consumers expect a response within 24 hours—and 38% expect a response within 4 hours, according to Klaviyo’s 2025 Future of Consumer Marketing Report. The more customers get used to fast, round-the-clock answers, the less patience they have for a brand that goes dark after 6 p.m.

Meeting that expectation used to mean staffing round-the-clock teams in every market. Now, it means AI that speaks the customer's language on the customer's channel, whenever and wherever they show up. An AI customer agent is how you make sure the French person sending you a WhatsApp message at midnight gets the same answer, at the same quality, as the American who starts a web chat at noon.

These conversations also need to happen on the channels customers actually use: web chat, email, texting, WhatsApp, and Instagram. Ideally, if a customer reaches out on one channel and later follows up on another, both AI and human agents have instant access to those threads so nobody ever loses context or has to repeat themselves.

6. AI knows when to escalate and hands off with full context

Good customer service AI is defined as much by what it resolves on its own as by what it routes to a person. A sensitive billing dispute or a tangled multi-order problem should go to a human, fast. What separates a good handoff from a frustrating one is whether the customer has to start over.

Frankly, they shouldn't have to. When AI escalates a ticket, the helpdesk inbox should open with the complete conversation history, across channels, alongside profile data, open orders, loyalty status, and the AI agent’s notes already loaded. The human picks up exactly where the AI left off.

When tea brand Harney & Sons’ AI customer agent detects that a customer enquiry requires a human touch, it passes the conversation on via the brand’s AI-powered helpdesk with full context preserved. The support team can now resolve tickets with less toggling between tabs, contributing to wins like a 25% PoP reduction in average service ticket resolution time in 30 days.

Some customer service questions are a natural fit for AI, while others still need a person. The line between them is more practical than you might expect:

7. AI turns support conversations into marketing and retention data

Every support conversation is also data, and too often, it just kind of disappears. Someone tells your team they're shopping for a gift, asks about a fragrance-free option, or mentions they're restocking a favourite, and that intent and preference, stated in their own words, usually vanishes in a closed ticket.

In an ideal world, your AI feeds that intent and sentiment back into the customer’s profile and keeps it working for you in the future. For example, a web chat resolution might capture a product preference, update the profile, and trigger a personalised follow-up a few days later, so the next marketing touch reflects what the customer just said.

That's a service-to-marketing loop where zero-party data from support interactions feeds segmentation and flows, and it’s how customer service stops being a cost centre and starts driving revenue and retention.

'Service is a part of retention, and it's part of marketing', Thorngren points out. 'Having our data centralised in one autonomous B2C CRM helps us have the real conversations, instead of giving people one-size-fits-all answers'.

Build out your own AI customer service examples with Klaviyo

With Klaviyo B2C CRM, all of the following capabilities live in one system, so your team can resolve customer issues, ship support-informed campaigns and flows, and personalise across the lifecycle without waiting on syncs or engineering:

• Data: Klaviyo Data Platform maintains a real-time, lifetime view of each customer, fed by 350+ pre-built integrations. Everything else in Klaviyo runs on these profiles.
• Intelligence: Klaviyo AI (K:AI) decides the best timing, channel and content for each individual person across email, text messaging, mobile push, WhatsApp. Predicted lifetime value, churn and next purchase date tell you what each customer is likely to do next, so you can act at the right moment instead of reacting after the fact.
• Agentic AI for customer service: K:AI Customer Agent works 24/7 across web chat, email, text messaging, and WhatsApp. When it can’t solve a problem, it escalates to your human team in Klaviyo Helpdesk with full context preserved.
• Personalised on-site self-service: Customer Hub gives logged-in customers a personalised portal for order tracking, loyalty redemption, subscriptions, and reorders, along with AI-powered personalised product recommendations that drive revenue.
• Agentic AI for marketing: Klaviyo Composer builds a launch-ready campaign or flow from a plain-language brief, with audience and messaging optimised across channels. Nothing sends without your approval.
• Analytics: Klaviyo Analytics tracks impact across email, text messaging, push, WhatsApp, and non-Klaviyo channels so you can see what’s working and where AI is paying off.