Live Chat Staffing Calculator

Voice staffing uses Erlang C — a queuing model that assumes each agent handles one call at a time and callers wait in a real-time queue. Live chat breaks both assumptions: agents handle multiple simultaneous conversations, and customers tolerate short delays between messages rather than waiting on hold.

Little's Law and the offered load

The correct approach starts with offered load: the average number of sessions in progress at any moment.

This is Little's Law applied to a chat channel. If you have 30 sessions per 30-minute interval, each lasting 8 minutes on average, your offered load is:

From offered load to seated agents

With a max concurrency of 3 sessions per agent, you need at least ⌈8 ÷ 3⌉ = 3 seated agents.

But this assumes perfect utilisation. Real teams need an occupancy guardrail — a ceiling on how busy any agent should be. If occupancy would exceed your target (typically 80–90%), the calculator adds agents until it falls back within range.

Occupancy in chat vs voice

Chat occupancy is defined as:

Unlike voice, agents aren't "idle between calls" when occupancy drops — they're between messages, still monitoring active threads. Overstaffing chat has lower productivity cost than understaffing, which causes burnout and missed first-response SLA.

The concurrency setting

Industry practice varies: simple product support allows 4–6 simultaneous chats; complex technical or financial support typically caps at 1–2. Setting it too high increases first-response latency even if throughput holds up.

Shrinkage and scheduled headcount

Seated agents are the minimum needed to handle demand. Scheduled agents account for the fact that not every scheduled person is available at all times — breaks, coaching, off-phone tasks, training. A typical shrinkage factor is 20–30%.

First-response SL — directional only

Unlike voice, there is no closed-form SL formula for multi-session chat queuing. The calculator provides a Poisson slot approximation: it estimates the probability that a new session finds an available slot within your target time. Use it as a directional signal, not a guarantee.

For precise SL modelling of chat, you need simulation — Turnella's planning module handles this as part of a full workstream.