Conversational commerce: a complete guide for UK retailers

Conversational commerce lets shoppers discover products, ask questions, compare options, buy and get support through a single conversation, in a website chat, on WhatsApp, through a voice assistant or on a social platform.

That sets it apart from a basic chatbot answering a fixed list of questions. Conversational commerce connects to live product, pricing, availability and order data, so a shopper can check stock, narrow down options, place an order and track it later without starting over each time.

WhatsApp alone reaches 90% of UK online adults, with 74% using it daily, according to Ofcom's Online Nation 2025 report, a sign of how embedded conversational channels already are in UK life. AI shopping assistants are moving just as fast: Adyen found UK use of AI shopping assistants climbing from 12% to 28% within a year, with 44% of shoppers willing to hand AI the whole purchase, from setting preferences through to paying, once those preferences are locked in.

AI assistants are one part of a wider category that also includes live chat, messaging, SMS and voice, which is why market-size estimates vary so much depending on what each researcher counts. Looking at how individual channels are actually used in the UK gives a clearer picture than any single market figure could.

This guide covers what conversational commerce is, the channels it spans, how businesses are using it, its benefits, UK examples, how to choose a platform, implementation, and UK compliance requirements.

What is conversational commerce?

Conversational commerce is the use of chat, messaging or voice to help people shop. Instead of searching through a website on their own, a shopper can ask questions, get recommendations, compare products, make a purchase or get help with an existing order through a conversation.

For example, say someone has been looking at a dress online. She opens the retailer's chat and asks, "Do you have this in a size 12?"

The system knows which dress she means, checks live stock and confirms it's available. She then asks, "Can I get it by Friday?" It checks the delivery cut-off for her postcode and confirms the date. If it works, she can complete the purchase from the same conversation.

What makes this conversational commerce is that the conversation is part of the shopping process itself. It remembers what the shopper is talking about, uses the retailer's product, inventory and order data to respond, and can help her take the next step.

That next step doesn't have to be a purchase. A shopper might use the conversation to compare two laptops, ask about returning something or check when an order will arrive.

Conversational commerce vs chatbots, social commerce and agentic commerce

Conversational commerce gets compared to a few other terms. Customer-service chat is a genuine comparison, as both happen through a direct conversation with a retailer. Chatbots, social commerce and agentic commerce describe different parts of the experience, so it helps to look at each separately.

For example, someone asking, "Where's my order?" wants an update on a purchase they've already made. Customer-service chat can check the tracking details and answer the question. But "I bought a jacket from you in October. Do you have something similar for winter?" is asking for help with a new purchase. There's no single answer to pull up. The system has to use what it knows about the previous purchase and what the retailer sells now to make a useful recommendation.

A chatbot is the interface, or the chat window itself. It might only answer FAQs and never touch a purchase, or it might connect to the retailer's catalogue, stock and checkout and become the way conversational commerce is delivered to the shopper.

Social commerce is about where a purchase happens rather than how the conversation works. Someone can see an outfit on Instagram, tap the product and buy it within the app, with no conversation involved. If they message the brand about sizing first and buy through that exchange, the same purchase counts as both social commerce and conversational commerce.

Agentic commerce is about how much control the shopper hands over. A customer might tell an AI agent, "Reorder my usual coffee whenever the price drops below £8," and leave the agent to monitor the price and place the order on its own once it does. Conversational commerce usually keeps the shopper involved in the final decision; agentic commerce can remove that step.

How does conversational commerce work?

Conversational commerce has to do four things: understand what the shopper wants, check the right information, take action, and know when a person needs to step in.

If someone says, "The coat I ordered came in a medium, I ordered a small," the system needs to understand that there's a problem with an existing order. It can then check the order, see whether a small is in stock, and work out what can be done.

The information it checks depends on the request. That could include order details, stock, prices, delivery updates, or the returns policy. Once it has what it needs, it can arrange an exchange, confirm a return, update a delivery date, or suggest another product.

It can also remember the conversation, so if the shopper comes back a few days later, they can continue from where they left off instead of starting again.

Some requests still require human intervention, especially when they involve an exception or a decision that falls outside the usual policy. In those cases, the system can pass on the conversation and order details so the shopper doesn't have to repeat everything.

This depends entirely on the data being current. Old stock figures, prices, policies or order details produce a wrong answer, delivered with the same confidence as a correct one.

Which channels does conversational commerce cover?

Conversational commerce spans websites and apps, messaging services, social media, voice and video, but each channel tends to suit a different kind of conversation. Knowing which is which helps a retailer decide where to start.

1. Website and app chat work well when a shopper is already browsing or managing an order. They can ask about a product, check a delivery or get help with a return without leaving the retailer's website or app.

2. WhatsApp is better suited to conversations that continue over time, from questions before a purchase to support afterwards. It reached 90% of UK online adults in May 2025, with 74% using it daily, according to Ofcom. A CM.com audit of the UK's top 25 retailers, reported by InternetRetailing, also found that retailers using WhatsApp scored 58% higher for response speed and effectiveness. Payments are still limited within WhatsApp in the UK, however, so shoppers will usually need to follow a secure checkout link to complete a purchase.

3. SMS and RCS are more useful for short updates such as delivery notifications, reminders and confirmations. Google Messages, which supports RCS, reached 20.4 million UK users in 2025, up 61% year on year according to Ofcom, making it an increasingly important way for retailers to reach customers.

4. Instagram, Messenger and other social platforms can turn interest in a post or product into a conversation straight away. Someone can see an item and ask about its size, price or availability without leaving the platform. The same CM.com audit found that almost half of unanswered customer queries came through Instagram, so these channels need proper support to be worth opening at all.

5. Voice and video work best when a shopper needs more help before deciding, particularly for more expensive purchases, consultations or accessibility needs where speaking to someone or seeing the product can be more useful than text.

6. Third-party AI assistants shift the retailer's role entirely. Instead of running the conversation itself, the retailer supplies product data to an AI platform that decides how to describe and recommend items. John Lewis has announced plans to make its products discoverable this way, through ChatGPT and Gemini, with purchasing to follow when the technology becomes available in the UK. The retailer's job is to ensure that data is accurate and complete enough for someone else's system to represent it correctly.

Conversational commerce use cases across the retail journey

A retailer can put a conversation to work at any stage of the shopping journey, not just at checkout.

1. Discovery and consideration

   1. Letting a shopper describe what they need instead of relying on them to search by the right product name

   2. Surfacing two or three options for direct comparison

   3. Confirming size, fit or compatibility before an order is placed, rather than after it's returned

   4. Checking live stock, in a specific store or online

   5. Recommending a product based on what someone's bought before

2. Conversion

   1. Building a basket inside the conversation itself, without a separate site visit

   2. Explaining exactly how a promotion or discount applies, before confusion turns into an abandoned basket

   3. Following up with a shopper who's left items unpurchased

   4. Handing off to a secure checkout once a shopper is ready to pay

   5. Booking an appointment or consultation ahead of a bigger purchase

3. Post-purchase

   1. Handling order tracking and delivery changes without a support ticket

   2. Managing an exchange or warranty claim

   3. Processing a return or refund request

   4. Prompting a reorder at the right moment

   5. Following up with a loyalty offer or a relevant recommendation

The conversation has to stay within both the shopper's statutory rights and the retailer's returns policy. A mistake here can mean giving a refund when it isn't due, or refusing one when it is.

What are the benefits of conversational commerce?

Conversational commerce can improve a few specific parts of the shopping experience. The important thing is to measure whether it's actually helping, rather than how many people are using the chat.

1. Higher conversion: Shoppers often have questions before they're ready to buy. If they can get a quick, accurate answer, they're more likely to complete the purchase. Retailers can measure this through conversion rates and the time between a shopper's first question and checkout.

2. Better recommendations: Conversations can make recommendations based on what someone has browsed, bought, or asked about. That makes the suggestions more relevant and can increase average order value. If the recommendations are too generic, they're unlikely to make much difference.

3. Lower service costs: Many customer questions are straightforward, such as whether something is in stock or when an order will arrive. If the conversation can handle these without an agent, customers get answers faster and service teams spend less time on routine requests.

4. Better human handoffs: Not every issue can or should be handled automatically. When an agent needs to step in, the resolution rate of escalated conversations is the number that matters, since it shows whether the agent can actually solve the problem. The conversation should give them enough context to continue from where it left off, rather than making the customer start again.

None of these benefits are guaranteed. If the conversation relies on outdated information or can only deal with a limited set of questions, it can create more problems than it solves. That's why retailers should also track things like unanswered questions, complaints, opt-outs, and incorrect refund decisions.

What to look for in a conversational commerce platform

Here are a few things to consider before choosing a platform:

• Covers the channels shoppers actually use: website, app, WhatsApp, voice

• Pulls stock, pricing and order data live, not from a copy that updates once a day

• Connects properly to the retailer's existing CRM and order management systems

• Can complete an action itself: build a basket, process a return, confirm an exchange

• Hands off to a secure checkout that meets UK payment rules

• Won't let the AI promise a discount, delivery date or exception it isn't authorised to give

• Passes the full conversation to a human agent, not just a case number

• Handles customer consent properly, and deletes data when it's supposed to

• Reports on whether conversations actually led to a sale or resolved a problem

• Works in more than one language, and for a screen reader, if that matters to the retailer's customers

• Logs every action taken, so a refund or exchange can be traced back later

• Costs more than just the monthly licence, so ask about setup and upkeep before signing

Most vendors here started in one of four places, messaging apps, customer service tools, ecommerce platforms, or AI built for shopping, and that background defines their strengths today.

A few retailers are already doing parts of this well. AJIO uses Fynd's ZIP assistant to let shoppers describe what they're looking for in their own words, rather than searching by product name. Chedraui uses WhatsApp Business to keep customers updated after a purchase and to collect feedback once an order has arrived.

How to implement conversational commerce?

The steps below work best in order, starting small and expanding once each stage holds up.

Step 1: Pick one journey to start with, a sizing question, a return request, something customers already ask about often. Prove it works there before adding anything else.

Step 2: Record where things stand before changing anything: how long this currently takes, how often it resolves cleanly. That gives a baseline to measure against later.

Step 3: Get the data right first, stock, pricing, order history, the returns policy that actually applies. This step takes longer than it looks, and rushing it causes most of the problems that show up afterwards.

Step 4: Build around the channel customers already use for this. If most queries already arrive by email, start there.

Step 5: Set the boundary clearly, what the system can answer or do on its own, and where it hands over to a person. Write the rule down so the team has something concrete to check the system against.

Step 6: When a person takes over, give them the full conversation. The shopper shouldn't have to repeat themselves.

Step 7: Test the edge cases: an unusual request, a rude message, someone using a screen reader.

Step 8: Pilot it with real customers before a full rollout. Real customers ask things a script never accounted for.

Step 9: Once live, watch where conversations fail. That's a clearer signal than the total number handled.

Step 10: Add a second journey once the first is solid. Moving too fast here means fixing problems in two places at once.

Here are some mistakes that tend to repeat across rollouts:

• Treating a basic FAQ bot as conversational commerce

• Running it on outdated stock or pricing data

• Automating refunds that ignore the customer's actual statutory rights

• Sending marketing messages with no consent in place

• Letting the AI pretend to be a person

• Dropping a shopper's context the moment a human takes over

• Tracking volume alone, without checking whether conversations actually resolved anything

• Switching on every channel before proving one works

UK compliance, privacy and trust checklist

The CMA published guidance in March 2026 on how consumer law applies when a business uses AI agents to talk to customers.

The core point: consumer law applies the same way whether a customer talks to a person or an AI, and a retailer stays responsible for what its AI says and does, even if a third-party vendor built it. A retailer should disclose when a customer is interacting with AI if not doing so could influence their decision or mislead them. The AI must also accurately represent its capabilities and avoid making claims about what it can do that go beyond its actual abilities.

The obligations themselves come from existing law, mainly the Consumer Rights Act 2015 and the Consumer Contracts Regulations 2013, covering accurate information, cancellation rights and fair terms. The CMA's power to fine, separate from those laws, comes from the Digital Markets, Competition and Consumers Act 2024, and can reach up to 10% of global turnover.

A few other requirements apply on top of this:

• UK GDPR covers any personal data collected or used during the conversation

• PECR requires consent before sending marketing messages, with a working opt-out

• Pricing shown through the conversation needs to be clear and complete

• Strong Customer Authentication applies to payments taken this way, same as any other channel

• A record of what the AI said and did should be kept, so an exchange or refund can be traced back later

• A person should always be reachable if the customer needs one

This isn't legal advice. A retailer's own legal team should confirm how this applies to their specific setup.

Getting started

Conversational commerce is a natural, data-connected way to help a shopper move through a real part of their journey, a question before buying, a problem after.

Start with one journey worth solving, prove it works, and expand from there once it holds up.

Fynd Kaily brings this together in one place, across website, WhatsApp, app and voice, with the full conversation history carried over whenever a person needs to step in. 

Find your first conversation to automate. 

Explore Fynd Kaily