WFM guideMultichannel

Response time targets across channels

A voice "service level", an email "SLA", and a chat "first response time" are three different things measured three different ways. Treating them as one number is how multi-channel operations set incoherent targets — and apply the wrong staffing model to the deferred channels.

The target reflects how the customer experiences the wait

Each channel's service metric exists to capture how the customer feels the wait. On voice, the customer is holding the line — every second is felt, so the metric counts seconds. On chat, the customer is present but more tolerant, and agents handle several conversations at once — so the metric is responsiveness, not answer-speed. On email and case work, the customer is absent and will check back later — so the metric is a deadline in hours or days, not seconds. Because the interaction modes differ, the metrics differ, and crucially the staffing models differ too. You cannot run an email channel to a voice-style service level, and you cannot collapse all channels into one target without losing meaning.

Service targets by channel

Voice

Service Level (SL)

Typical target

80% answered in 20s (varies — sales often higher, e.g. 90/10)

How it's measured

% of calls answered within the threshold, measured per interval. The customer is live on the line, so seconds matter.

Staffing model

Erlang C — staff to answer random live arrivals fast.

Live chat

First Response Time (FRT) + ongoing response time

Typical target

First response in 30–60s; ongoing replies within ~1–2 min

How it's measured

Time to the first agent message after the customer opens the chat, plus reply latency within the conversation. Concurrency means an agent juggles several chats, so 'wait' is more elastic than voice.

Staffing model

Concurrency-adjusted model — accounts for agents handling multiple simultaneous conversations.

Email / ticket

Response & resolution SLA

Typical target

First response in 1–24h; resolution within 1–3 working days

How it's measured

Time from receipt to first response, and to resolution, against an SLA deadline. The customer is not present — they will check back later.

Staffing model

Backlog / flow model — staff to clear inflow plus backlog before deadlines.

Back-office case work

Case SLA deadline

Typical target

Varies widely by case type — days to weeks (e.g. complaints 8 weeks)

How it's measured

Time from case creation to completion against the per-case-type SLA. Prioritised by breach risk, not arrival order.

Staffing model

Backlog / flow model with breach-risk prioritisation.

Setting a coherent multi-channel target framework

Set each channel's target in its own metric

Voice in SL (% in X seconds), chat in FRT (seconds/minutes to first response), email and case work in SLA deadlines (hours/days). Don't force one metric across all channels — a single 'service level %' that spans voice and email is meaningless.

Make the targets consistent in customer intent, not in number

The targets should express a consistent service ambition across channels (e.g. 'fast and responsive everywhere'), even though the numbers differ. A 20-second voice SL and a 4-hour email SLA can both express 'good service' for their channel — they are not meant to match.

Staff each channel with its own model

Voice on Erlang C, chat on the concurrency model, email and case work on the backlog/flow model. Reporting can roll up to a combined view, but the underlying requirement must be computed per channel with the right model — never a single blended Erlang C.

Beware blended agents splitting attention

When agents handle multiple channels, the real-time channel (voice/chat) almost always wins their attention, and the deferred channel (email/case) silently slips against its SLA. If agents are blended, protect the deferred-channel SLA explicitly — ring-fence time or it will be the channel that quietly breaches.

Response time target questions

Why is service level measured differently for voice, chat, and email?

Because the customer experiences the wait differently in each channel. On voice, the customer holds the line live, so the metric is service level — % answered within a few seconds (e.g. 80/20). On chat, the customer is present but more tolerant and agents handle several conversations at once, so the metric is first response time plus ongoing reply latency, not a single answer-speed figure. On email and case work, the customer is absent and will check back later, so the metric is an SLA deadline in hours or days. These aren't interchangeable — voice SL optimises for fast answer under random live arrivals (Erlang C) while an email SLA optimises for clearing a backlog before deadlines (a flow model). A single cross-channel 'service level' number, or a voice-style target applied to email, produces incoherent goals and the wrong staffing model for the deferred channels.