Conversational commerce: What it is and how it works

Conversational commerce: What it is and how it works

Conversational commerce is where buying, selling, and customer service is carried out inside a two-way conversation on messaging apps, text, in-app chat, or voice, where the shopper can ask, decide, and complete a purchase or a service request without leaving the thread.

These conversations can be handled by a person or an AI agent, and most customer engagement programs use both. An agent typically handles order status questions and routine reorders, and a person takes over when a thread needs judgment.

This article explains what conversational commerce is, how it works, and the channels it runs on. If it's specifically the AI component behind these conversations you're after, check out our conversational AI in eCommerce guide.

TL;DR

• Conversational commerce lets shoppers buy and get help inside a two-way message thread, with people, AI agents, or both answering on the brand's side.
• A conversation only works when the brand has the shopper's consent to message them, recognize them, and see what they’ve purchased.
• WhatsApp, SMS and RCS, and in-app messaging give brands conversations they own, while social messaging and voice add reach but leave most of the data with the platform.
• Benefits like higher conversion and more repurchases depend on conditions, such as live stock data, a clear escalation route, and replies that are saved back to the customer profile.
• AI makes conversations more contextual and lets agents act on requests. The best engagement platforms keep conversations and campaigns on a single customer profile, with results measured against a control group.

What is conversational commerce?

Conversational commerce is where buying, selling, and customer service are carried out inside a two-way conversation on messaging apps, text, in-app chat, or voice, where the shopper can ask, decide, and complete a purchase or a service request without leaving the thread.

What makes it commerce and not just a support chat is where and how far the conversation can go. Support chats answer a question and close a ticket. A conversational commerce thread can move the shopper forward with a product recommendation, a size confirmation, a payment link, a reorder, or a return, handled with engaging replies that don't feel like a standard bot response.

The outcome is a completed transaction or a customer retained, which goes well beyond a closed ticket and impacts revenue.

Two things separate conversational commerce from its predecessor, a contact form with a chat window. The first is that the conversation is two-way. The shopper can reply, ask a follow-up question, or change their mind, and the brand responds within the same thread.

The second is that the conversation has context. The brand knows who the customer is and what they've bought before. That way, "Where's my order?" gets an answer about that specific order, and "Same as last time, please" can become a reorder.

Conversational marketing is different in that it starts and nurtures the conversation, whereas commerce completes the transaction.

How conversational commerce works

Consider a shopper who sends an SMS message to a skincare brand, asking: "Can I get my usual moisturizer again?" Behind the scenes, four things have to happen before that message becomes an order.

1. The shopper gives permission or starts the thread

When the shopper sends the first message, they've given the brand permission to reply. A brand can only start the conversation, say, with a restock reminder, if the shopper has already given explicit consent to hear from it on that channel. That consent usually comes from a checkbox at checkout or a sign-up form on the brand's website, and each channel sets its own rules, from WhatsApp and RCS to SMS in the US. Keeping each shopper's opt-in status by channel on their customer record tells the brand who it can message, and where.

2. The message arrives with the shopper's identity attached

The number the text came from matches a customer record, so the brand knows this shopper has bought the same moisturizer three times since spring. Because every reply in a two-way conversation comes back into the same thread, the brand never loses track of who it's talking to. Anything new the shopper shares along the way, such as a change of skin type or a preferred delivery day, is added to that same profile as first-party data.

3. A person or an AI agent replies with context

The person or AI agent who replies needs context, meaning what the brand already knows about the shopper. In this case, that's their last order and whether the moisturizer is in stock. Without it, the person or agent would need to ask for the product name, an order number, or other details, making the shopper do the work the conversation was supposed to save them.

4. The request completes in the thread or moves on

In this case, the agent replies with the moisturizer, the price, and a button to confirm. One tap and the order goes through without the shopper leaving the thread, but not all requests will close that neatly. A shopper considering a new serum might need a checkout link with the item already in the cart, and one asking about a skin reaction needs a human handoff. Once a person takes over, they can read everything that’s been said so far, so the shopper doesn't have to repeat themselves.

Conversational commerce examples and use cases by channel

Each channel shapes the conversation differently, from how rich the replies can be to who owns the data, what permission you need, and what each message costs. Here's how the five main channels compare, with an example of each in action, covering the most common conversational commerce use cases.

WhatsApp

WhatsApp supports rich two-way threads with product carousels, quick reply buttons, and payment, and it's the default messaging app in many markets. Our WhatsApp marketing guide covers the formats in detail.

Example: A shopper messages a fashion retailer to ask, "Do you have the green linen dress in a 12? Need it for Saturday." The retailer replies with a carousel showing the dress in her size and the delivery cutoff for Saturday. She taps the card, pays in the thread, and gets her order confirmation in the same chat.

The channel works for more than purchases, though. Stori, a fintech company serving Latin America's underbanked population, used two-way WhatsApp messaging to ask customers for updated address details, which customers could update in one click within the message. Stori reported 26% higher conversion than its in-app process among users updating their address.

Consent and trade-offs: Shoppers need to opt in, ideally to WhatsApp specifically, before a brand can message them first. Because shoppers opt in with their phone number, the brand can tie each WhatsApp conversation to its customer record. Replies to a shopper are free, but messages the brand starts use Meta-approved templates, which have been charged per message since July 1, 2025. In the US, marketing messages are paused, so brands are limited to service, utility, and reply messages.

Best for: Brands selling into markets where WhatsApp is the default messaging app, and teams ready to manage templates and opt-ins on a dedicated channel.

SMS and RCS

SMS reaches almost any mobile phone, which suits reorders and order-status updates, or anything that requires urgent attention. RCS adds rich cards and quick replies, suggested wording for ease, and when a device can't receive RCS, there is a fallback to SMS.

Example: Aeroflow Health is a healthcare consumer-goods business supplying breast pumps, and its Mom & Baby division sends customers reorder prompts via SMS. When a customer replies that she needs a different size, an AI agent asks which size she'd like, finds the matching product in the catalog, and completes the order in the text thread. Aeroflow Health reported an 8.8% lift in repurchase conversions over business as usual.

Consent and trade-offs: The brand holds the shopper’s phone number and the full text thread, so each reply can feed straight back into the customer profile. In the US, CTIA's guidelines expect express permission for informational texts and express written permission for promotional ones, and RCS requires clear, affirmative consent. SMS is plain text, and every message has a price. Our guide to SMS marketing goes into more detail.

Best for: Replenishment and subscription brands, and brands adding their first two-way channel, since SMS reaches almost any phone without an app or account.

In-app and web messaging

Here, the shopper is logged in and already on the brand's app or site, so the conversation starts with their profile, cart, and order history in view.

Example: A shopper in a grocery app opens a chat and asks, "Can I swap Thursday's delivery to Friday morning?" The agent replies with the open Friday slots, the shopper picks one, and the order updates in the app.

Consent and trade-offs: The conversation happens inside the brand's own app or site, so there's no separate opt-in and no carrier fee per message. The trade-off is that shoppers have to download the app to use in-app chat, which limits it to a brand's more engaged customers.

Best for: Brands with high app usage or frequent logged-in visits, and those with an established app they want to turn into a sales and service channel.

Social commerce messaging and marketplace chat

Instagram DMs, Facebook Messenger, and marketplace chat reach shoppers who may never visit the brand's own site, but the platform owns the conversation.

Example: A shopper sends a DM about a jacket in a brand's Instagram post asking, "Is this waterproof or just water-resistant?" The brand replies, along with the product rating and a link, and the shopper purchases it on the brand's website.

Consent and trade-offs: The platform sets the rules, but data is the real cost. Social platforms typically identify shoppers with their own IDs and don't share contact details with the brand. On Facebook Messenger, for example, each shopper gets an ID specific to the brand's page. Unless that ID is matched to a customer record, the brand can't see the shopper's purchase history, so they may have to repeat details the brand already holds, and the conversation doesn't inform what they receive next.

Best for: Brands whose shoppers discover products on social platforms, and that already have a customer profile to match social conversations against.

Voice assistants

Some shoppers buy out loud, asking an assistant such as Alexa, Google Assistant, or Siri to place an order from a smart speaker, phone, or TV. Many of those devices have no screen, so voice suits shoppers who already know what they want, which today mostly means reorders.

Example: A shopper says, "Alexa, reorder the dog food." The assistant reads back the brand and pack size from the last order and confirms the saved payment method. The shopper says yes, and the order goes through.

Consent and trade-offs: The conversation and payment run through the assistant's own account, so the brand usually sees the order but not what the shopper said. Without a screen, comparing products is hard, and the assistant can only find the right item if the brand's product data with that retailer is complete and up to date.

Best for: Brands selling replenishable household products through retailers that support voice ordering, as an add-on to an established reorder program.

The benefits of conversational commerce in retail, and what to consider

Conversational commerce in retail returns four things: higher conversion on assisted decisions, lower cost per service request, more repurchases, and richer first-party data. Each depends on a condition explained below.

Higher conversion on assisted decisions

A quick answer in the thread can be what moves someone from browsing to buying. When someone asks whether a jacket runs small or comes in navy, a reply in the moment lets them decide while they're still interested. For that to work, the person or AI agent replying needs to see current stock and price. Otherwise, the recommendation may point to something the shopper can't buy.

Lower service cost

An AI agent can handle simple requests, such as resending a receipt or checking on a refund, without anyone on the service team stepping in. That lowers the cost of each request and gives the team more time for complex cases. The savings depend on shoppers being able to reach a person when the agent can't help. If they can't, the request gets abandoned, and sometimes the order goes with it.

More repurchases

Reorder reminders sent in a conversation let shoppers buy again with a single reply. For brands selling products people run out of, that can bring customers back on a regular cycle. The reminder needs the shopper's opt-in on that channel, and it needs to arrive around the time they're likely to run low.

Richer first-party data

In conversation, shoppers share details they might never enter in a form, like the size they usually wear or a dietary need. Saved to the customer profile, those details improve the recommendations and messages the brand sends on every channel. If they stay in the chat history, no other channel can use them.

How conversational AI fits in

AI does two jobs in conversational commerce. Selection decides who gets a message and when, such as which shoppers receive a reorder reminder this week. Generation writes the replies once the conversation is underway. So selection works before a conversation starts, and generation works inside it. Both are used well beyond messaging, across AI in eCommerce.

AI also makes conversations easier for shoppers. Instead of replying with a set keyword, a shopper can write naturally, like "the blue one but a size up," and AI understands what they mean. It also remembers what's already been said, so a later question like "Can it get here by Friday?" gets an answer about the right order. Through AI agents and live agents working inside channels like WhatsApp, SMS, and in-app messaging, it can act on requests too, such as placing an order or changing a delivery, within limits the brand sets. That's what makes conversational AI in eCommerce useful from a shopper's first question through to their reorder, provided the right data and guardrails are in place.

Adding AI doesn't remove the basics, though. Brands still need permission before they message a shopper, whether the message comes from AI or a person. AI also needs access to product and order information, like order status and reorders, to give correct answers. Without current stock levels, it might recommend a product that has already sold out or that the customer is already stocked up on. And some conversations still need a person. When a shopper is complaining about a late delivery for the third time, they need someone who can make things right.

How Braze supports conversational commerce

Conversational commerce works when each conversation knows who the shopper is and what has already happened with them. Braze runs conversations from the same customer profile as the rest of a brand's messaging.

That profile sits behind WhatsApp, SMS and RCS, and in-app messaging, which are built into Braze alongside email and push. So when a shopper replies in a WhatsApp thread, that reply can influence the next email or push notification they receive.

Inside those conversations, BrazeAI™ Agents handle two-way requests, working within the goals and guardrails the team sets. When a request needs a person, conversations on WhatsApp can be handed to live agents.

You can understand effectiveness because Braze reports the lift a conversation program produces by comparing customers who received the messages against a global control group that didn't.

What to look for in a conversational commerce platform

For an enterprise B2C brand, the right platform comes down to what each conversation can know and do, and whether the brand can prove the conversations are paying off. These five capabilities cover the essentials:

• Owned-channel coverage with consent built in. WhatsApp, SMS, RCS, and in-app messaging run from one place, with each shopper's opt-in status tracked by channel.
• A shared customer profile. Conversation channels and campaign channels like email and push work from one profile, so a WhatsApp exchange and next week's email draw on the same information.
• Access to catalog and order systems. Replies can look up inventory and order details, and complete actions like a reorder or a delivery change.
• AI agents with guardrails and clean handoff. The team sets what agents can say and do, and a person can take over a conversation with the full history in view.
• Measurement against a baseline. Reporting splits out conversation-assisted conversions from sales that would have come in anyway.

What a brand needs from a platform depends on how mature its conversational program is. For a brand opening its first two-way channel, such as SMS, the priority is the groundwork: a record of which shoppers have agreed to be texted, and replies that can draw on the purchase history the brand already holds.

A brand running conversations on WhatsApp, SMS, and in-app at once has a different challenge. Its shoppers move between those channels, so a question started in one needs to carry over to another without losing any context. AI agents should behave consistently wherever a shopper reaches them, and results need measuring across the whole program.

Where conversational commerce is going: conversational commerce trends

Conversational commerce is still changing, and a few trends in particular will affect brands building a program today.

AI agents on both sides of the conversation

Shoppers are beginning to use their own AI agents to research, compare, and complete purchases, which means a brand's agent may soon be talking to another agent. That puts more weight on product details and stock levels being complete and current, since a shopper's agent will rely on them to find the right product and place the right order.

RCS taking over from plain SMS

Since September 2024, iPhone Messages has supported RCS when people message non-Apple devices, giving them richer media and more reliable group messaging compared to SMS and MMS. Text programs can now include product images and reply buttons, with a text-only version for phones that can't show them.

Checkout inside a conversation

Paying within a messaging app is becoming normal, so a shopper can pick a product and pay for it without switching to a checkout page. For that to work, payment and order systems have to be linked to messaging, and the simplest time to link them is when the first channel launches.

Conversations and campaigns on one profile

Conversations have often run in a separate chat or support tool, apart from email and push. More brands are bringing both onto one platform, where an SMS exchange and a promotional email work from the same customer data. When choosing a platform, check that conversations and campaigns will share that data.

See how brands turn WhatsApp, SMS, and in-app conversations into orders with Braze.