Customer Service Messaging: A Complete Guide

customer service messaging

Customer service messaging lets people write to your team and get answers on their own time, instead of waiting in a live queue. Choose your conversation style first, then measure it with metrics that fit that style.

Item Value
Default “service level” threshold in Microsoft’s Omnichannel report (adjustable) 30 seconds
Default first-response SLA threshold in the same report (adjustable) 1 minute
Microsoft intraday insights reports deprecated as of May 1, 2023
Initial live wait in a Zendesk community example 60 seconds
Wait time the same report showed after a follow-up reply 14,430 seconds

The thresholds come from Microsoft Learn and the wait-time example from a Zendesk Community post. Vendor settings and reports change, so check the source before you rely on these figures.

What is customer service messaging?

Customer service messaging means handling support through digital messaging channels, not only through phone calls or a live-chat window. Microsoft describes its Omnichannel for Customer Service as a suite of capabilities that lets organizations instantly connect and engage with customers across digital messaging channels.

In Microsoft’s reporting, the digital messaging category covers social channels, SMS, Microsoft Teams, entity records and custom messaging channels. So “messaging” is a family of channels, not one app.

The biggest practical difference from classic live chat is that a conversation can be picked up again later. Zendesk’s migration guidance notes that messaging has no direct equivalent of a dropped chat, because conversations can be resumed later. That one fact changes how you count, staff and report.

Choosing the tool is a separate decision, and our guide to business messaging software covers that side. Promotional messaging is a different topic, covered in our guide to whatsapp marketing.

Pick a conversation style before you measure anything

Zendesk’s guide to moving from live chat to messaging says one of the main decisions is which conversation style you implement, because the style changes the metrics you use. It describes three styles.

Style What it is Effect on metrics
Live chat A single live interaction with a clear start and end Metrics compare directly with legacy live chat; Zendesk recommends it for an initial migration
Reopened Multiple live interactions; the ticket is solved, then the customer can return to it Expect inflated ticket-lifecycle metrics such as Requester Wait Time or Full Resolution Time
Asynchronous Multiple live and non-live interactions over a longer period Lifecycle metrics are less reliable for staffing or forecasting

The style descriptions and metric effects above come from Zendesk’s messaging metrics guide.

Higher numbers under the reopened style are not automatically a sign of worse service. Zendesk explains that a single reopened ticket would have been two separate chats in the legacy live chat experience.

If you are weighing the two approaches, read our comparison of live chat vs messaging customer support before you commit.

customer service messaging Support team
A man and a woman wearing headsets work at laptops beside each other.

Which metrics matter in customer service messaging?

Microsoft’s conversation insights report is a useful glossary of what a messaging dashboard can show. These are the entries most relevant to a support team, as the page defines them.

Metric What it tells you
Service level Share of chats accepted by an agent within the threshold
Longest wait time How long the longest-waiting open conversation has been in the queue
Average handle time Average time agents spent on sessions while they were in focus
Abandon rate Share of conversations where customers left before an agent accepted
Transfer rate by queue Share of sessions moved to another agent or queue
SLA – First response time Share of conversations that got a first agent response within the threshold

Both time thresholds are settings, not laws. The service level counts chats accepted within 30 seconds and the first-response SLA counts replies within 1 minute, and Microsoft says each can be modified to suit your business needs, per its Conversation insights documentation.

Read the fine print on what is counted. Microsoft’s handle-time, wait-time and SLA entries only consider conversations handled directly by an agent or escalated by a bot to an agent; conversations a bot handles and closes are excluded.

One caution on that source: Microsoft states that intraday insights reports are deprecated as of May 1, 2023 and recommends real-time analytics reports instead. Use the page for its metric definitions, not as a setup walkthrough.

For benchmarks, see our guide to the average customer support response time. For the wider scorecard beyond speed, our guide to customer service metrics and kpis goes further.

Carrying live chat numbers over to messaging

If you are migrating, Zendesk maps old chat metrics to messaging equivalents. A few of them:

  • Missed inbound chats become unreplied, unsolved messaging tickets, filtered by a status of New.
  • Transferred chats become reassigned messaging tickets, meaning tickets assigned to more than one agent.
  • Offline messages become messaging tickets created outside the schedule assigned when the ticket was created.
  • Chat satisfaction score becomes the count of messaging tickets with a good satisfaction rating.
  • Chat first reply time becomes first reply time: the gap between ticket creation and the first agent reply.

Why blending live and asynchronous data misleads you

A July 2023 post in the Zendesk Community shows the trap clearly. A customer contacted the team through live messaging and got an initial wait of 60 seconds, then sent an asynchronous follow-up that an agent answered in the next shift, about 8 hours later. The report averaged both and showed a wait time of 14,430 seconds, which the author said could make managers think agents were slow during the live exchange. The post is on the Zendesk Community.

As of that July 2023 post, the Zendesk product support team had no way around this. Community replies later in the year suggested leaning on SLA metrics, such as SLA First Reply Time in Business Hours, which should give a more accurate result for async communication.

The takeaway is practical:

  1. Decide which conversations are meant to be live and which are meant to be asynchronous.
  2. Judge live replies with a first-response or SLA measure, not a lifecycle average.
  3. Track “stale” conversations separately: Zendesk suggests looking at messaging tickets that have sat in a Pending status for longer than a day or week.
customer service messaging Headset closeup
A smiling woman holds the microphone of her headset near her mouth.

Build the daily workflow around replies

Once the style and metrics are set, the day-to-day work is about who answers, how fast, and with what wording. These guides each cover one piece of that workflow:

Automation changes your numbers too. Because Microsoft’s report leaves out conversations a bot handles and closes, you need to decide how to count that work, and our guide to chatbot vs human customer support helps frame that choice.

Handoffs deserve a watch. Transfer rate by queue shows the share of sessions moved to another agent or queue, and it includes sessions a bot created and escalated to an agent.

customer service messaging Office agents
Two headset-wearing colleagues sit behind laptops in an office.

Frequently asked questions

How is messaging different from live chat?

Show answer

In Zendesk’s terms, the “live chat” style is a single live interaction with a clear start and end. Messaging more broadly lets a conversation be resumed later, so it can include reopened or asynchronous exchanges.

What response time threshold should I set?

Show answer

Microsoft’s report uses 30 seconds for chats accepted by an agent and 1 minute for first response, and says both can be modified to suit your business needs. Treat them as starting points, not an industry rule.

Why do my messaging response times look worse than my chat times?

Show answer

Reopened and asynchronous conversations stretch ticket-lifecycle metrics such as Requester Wait Time and Full Resolution Time. A Zendesk Community example shows a blended report producing a 14,430-second wait from a 60-second live reply and a follow-up answered about 8 hours later.

Do bot-handled conversations show up in these reports?

Show answer

In Microsoft’s conversation insights definitions, conversations handled and closed by a bot are excluded from handle time, wait time and SLA figures. Conversations a bot escalates to an agent are included.