How to share customer context across chat, SMS, and voice

How to share customer context across chat, SMS, and voice

Customers aren’t thinking about your communications stack as much as you are, and that’s probably a good thing. A customer might start in your web chat about a delayed order, but when the chat can't fix it, they give your team a call. The agent who answers asks for their order number, email, and what happened. And…the customer has to explain everything again (but a little less patiently this time).

That's the gap omnichannel customer service is supposed to close, and most teams haven't closed it. In Twilio's Navigating the Data Deluge research, 76% of consumers said the agent has no or very little context about their issue, so they end up repeating themselves again. And that lines up with the fact that business leaders named connecting customer data across channels like voice, email, and messaging as their biggest obstacle to understanding customers (48%).

Fortunately, connecting your customer data across channels doesn’t have to be a massive, intimidating task on your to-do list. Here's how to share customer context across chat, SMS, and voice, so it follows the customer wherever they go, with Twilio.

What is omnichannel customer service?

Omnichannel customer service is support that connects every channel a customer uses, like chat, SMS, WhatsApp, and voice. This lets their history and context follow them from one channel to the next.

Multichannel support offers several channels, but that’s just the start. You need to connect them, and that’s where omnichannel comes in. A customer can text on Monday, call on Tuesday, and pick up where they left off, because every channel reads from the same customer record.

Broken context when switching channels is the #1 AI behavior hurting your customers' experiences, and we’re going to walk you through solving it here.

How to share customer context across chat, SMS, and voice

We’ll be using the following Twilio tools to share context across channels:

• Twilio Conversation Memory: Stores customer profiles

• Conversation Orchestrator: Ties conversations on every channel to those profiles

• Twilio Agent Connect: The open source SDK that connects your AI agent to Twilio's channels.

1. Tie every channel to one customer profile

You can only share context about a customer when you know it’s the same person. Conversation Memory's identity resolution maps each channel's identifier to a customer profile:

• Phone numbers for voice and SMS

• WhatsApp handles

• Chat IDs

• Email addresses

• Your own user IDs

Phone numbers are normalized before matching, so a number entered as 1-317-555-6789 matches +13175556789.

Upload customer profiles with linked identifiers, like user ID, phone, email, and WhatsApp number, before interactions begin. The Segment Conversation Memory integration keeps mapped traits in sync as they change in Segment. You’ll also want to link your customers’ second numbers. A customer who texts from one number and calls from another only unifies if both numbers are on the same profile.

2. Group conversations by customer profile

Conversation Orchestrator looks up each customer's profile by their address and links the conversation to it. Twilio's Orchestrator profiles docs walk through what this looks like in action: an SMS from a number on day 1 creates a profile, and a voice call from the same number on day 2 resolves to that same profile. 

With GROUP_BY_PROFILE grouping, the call lands in the same conversation as the text.

Set that grouping explicitly. Twilio's docs call it preferred for production, and it only works when profile resolution succeeds. If Orchestrator can't match a customer to a profile, it groups their conversations by phone number or address alone, so they won't connect to the customer's history on other channels.

3. Connect your channels

Conversation Orchestrator supports voice, SMS, RCS, and WhatsApp. Chat comes in through the Conversations API (classic) bridge, where each chat stays its own conversation, so context reaches it through the customer's profile.

On phone calls, ConversationRelay handles speech-to-text and text-to-speech over a WebSocket connection, with median latency under 0.5 seconds in Twilio's internal benchmarks. Your AI agent works with text on every channel (voice included).

4. Give your AI agent the same context everywhere

Agent Connect routes Voice, SMS, WhatsApp, RCS, and Chat through a single message handler, so one agent with one set of tools and rules serves every channel. Set the memory mode to always or once on each channel, and Agent Connect retrieves the customer's observations, summaries, and communication history before each message reaches your code. 

The default is never, so set it explicitly.

Our walkthrough on building a multi-channel AI agent with Twilio Conversations shows the pattern with a different agent framework—a customer's SMS and WhatsApp messages merge into one thread, and their stored facts load into the prompt on the first turn. 

For more on customer recall, see our guide to building an AI agent that remembers returning customers.

5. Carry context from chat to a phone call

When a customer leaves a web chat and calls, the call starts a new conversation, and the only bridge between the two is the customer's profile. This breaks a lot of omnichannel setups, but Twilio Conversation Memory fixes it. It adds observations and summaries to that profile after a conversation goes inactive or closes, so a customer who calls mid-chat can get there before the update does.

A couple of things help when a customer leaves a chat and calls right away:

• Save the facts as they happen: Give your chat agent a tool that writes key details, like the order number and the issue so far, to the profile as observations through the Memory API, without waiting for the chat to close.

• Extract on inactive: Twilio recommends adding extraction on INACTIVE when you need context available between interactions in under a minute.

Then make sure the call finds the same profile. If the customer's chat identity and phone number sit on one profile, identity resolution connects the two. If they don't, the agent on the call starts cold.

6. Hand off to people with the full picture

Context has to reach human agents, too. Agent Connect's escalation tool sends the conversation ID, memory store ID, and profile ID to your Studio flow, and in Twilio Flex, those become task attributes.

In Flex, Unified Profiles shows agents the customer's profile, interaction history, and traits before they accept a task (and that’s across voice, SMS, WhatsApp, email, and webchat). Agent Copilot adds AI-generated customer highlights. Both are in public beta. 

Twilio's Agent Connect and Flex handoff tutorial walks through the build, and our guide to designing an AI-to-human handoff that preserves context covers the methodology.

Let context follow the customer

Customers think in problems, and they expect the problem to follow them when they move from chat to a call. Tie every channel to one profile, group conversations by that profile, save what matters as it happens, and hand people the full picture. 

That's what omnichannel was supposed to mean all along.

Start for free and connect your channels with Conversation Orchestrator, or talk to sales about omnichannel support for your contact center.

Frequently asked questions

What is the best platform to share customer context between chat, SMS, and voice support?

Twilio Conversation Memory links each customer's phone, chat, and WhatsApp identifiers to one profile, and Conversation Orchestrator groups their conversations by that profile, so AI and human agents see the same context on every channel.

How do I pass full conversation context from a web chat to a live phone call?

Twilio Conversation Memory carries context through the customer's profile. Save key chat details as observations through the Memory API, and the agent on the call can recall them once the caller resolves to the same profile.

What is omnichannel customer service?

Omnichannel customer service connects every channel a customer uses, so their history follows them from chat to SMS to voice. Twilio does this by tying each channel's identifier to one customer profile.

What's the difference between omnichannel and multichannel customer service?

With Twilio, omnichannel means every channel reads from the same customer profile. Multichannel support offers several channels that each keep their own history, but that means customers repeat themselves when they switch.

How do you keep conversation history when a customer switches channels?

Twilio Conversation Orchestrator groups a customer's conversations by profile, so a text and a later call from the same number land in one conversation. Conversation Memory adds observations and summaries from past conversations.