WFM guideAsynchronous channels

Contact centre email and ticket WFM

Q: How do you calculate staffing requirements for an email queue?

Email staffing uses a throughput model, not Erlang C. The formula is: Agents required = (Expected new volume + Opening backlog) ÷ (Agent throughput rate per hour × Available hours). Example: if 800 emails are expected during a 10-hour operating day, the opening backlog is 120, each agent can handle 10 emails per hour (6-minute average handle time), and the target is to clear the queue by end of day: Total workload = 800 + 120 = 920 email-contacts. Total agent-hours required = 920 ÷ 10 = 92 agent-hours. Agents required for a 10-hour day = 92 ÷ 10 = 9.2, round up to 10 agents. This calculation must account for shrinkage (agents are not available for the full 10 hours), so divide by the productive percentage: 10 ÷ 0.75 = 13.3 FTE on-shift required. The throughput-based model should be applied at interval level (typically hourly) to account for variation in arrival rates and to identify when intraday backlog clearing may fail if volume is back-loaded.

Erlang C is the wrong model for email queues. Email is asynchronous — it accumulates as a backlog, not a real-time queue. The staffing model is a throughput calculation: how many agents are needed to clear expected volume plus opening backlog within the response time target?

Voice vs email: why the WFM model is different

Dimension	Voice / inbound phone	Email / ticket
Staffing model	Erlang C / Erlang A — queuing theory calculates minimum agents for real-time SL	Throughput model — calculates agents required to clear volume within a response time window
SL metric	% contacts answered within X seconds (e.g. 80% in 20 seconds)	% contacts responded to within Y hours (e.g. 90% within 4 hours or 80% within 24 hours)
Queue behaviour	Callers enter queue, wait, abandon or are answered. Queue clears in real time.	Emails accumulate as a backlog. Backlog is worked down over the operating day. No real-time abandonment.
Peak impact	Volume peaks create immediate queue spikes that require real-time response	Volume peaks create backlog growth. If agents cannot clear the backlog by end of day, it carries over to the next day.
Understaffing consequence	Immediate SL breach, long wait times, high abandonment in the affected interval	Backlog grows. Response time target is breached for contacts at the back of the queue. No customer abandons — they wait (silently).
Intraday management	Real-time SL monitoring; break timing; overflow routing; staffing adjustments	Backlog level monitoring; throughput rate monitoring; intraday queue clearing projections; redistribution of agents between email and other channels if needed
Handling time variability	AHT varies within a range; outliers are managed within the interval	Email AHT varies more widely than voice — a simple query and a complex complaint in the same queue may have very different handle times. Complex contacts must be managed separately or AHT modelling will be unstable.

The email throughput staffing calculation

Worked example — staffing a 10-hour email operating day:

Inputs

800 new emails expected today; 120 email backlog from yesterday; 6-minute average handle time (10 emails per agent-hour); target: clear queue by end of day; shrinkage rate: 25%

Step 1 — total workload

800 new + 120 backlog = 920 email-contacts

Step 2 — agent-hours required

920 contacts ÷ 10 contacts per agent-hour = 92 agent-hours of productive time

Step 3 — gross agents required (without shrinkage)

92 agent-hours ÷ 10 operating hours = 9.2 agents → round up to 10 agents on-phone

Step 4 — on-shift agents (with shrinkage)

10 on-phone agents ÷ (1 − 0.25 shrinkage) = 10 ÷ 0.75 = 13.3 → 14 agents on-shift

Step 5 — validate intraday

Apply arrival rate distribution by hour. If 40% of volume arrives in the first 3 hours (320 emails by 12:00) but only 4 agents are available for the morning, project the backlog position at each hour and confirm end-of-day clearing is still achievable.

This calculation must be repeated at hourly intervals intraday using actual arrival data and actual throughput rate rather than estimates. If backlog is growing faster than the intraday projection predicted, escalate before end of day — not after.

When carryover backlog is acceptable — and when it is not

Some operations accept a small carryover backlog (e.g. emails received in the last 2 hours of the operating day) and plan to clear them as first priority the following morning. This is operationally valid if the response time target is 24 hours and the carryover never exceeds the morning-shift processing capacity. It becomes a failure mode when the carryover grows night after night — each day starting deeper in deficit. WFM must track opening backlog position as a KPI and flag any sustained trend of growth.

Intraday management for email queues

Monitor throughput rate, not SL

The intraday metric for email is not SL — it is the hourly throughput rate (emails cleared per agent per hour) vs. the throughput required to meet the end-of-day clearing target. If throughput drops below the required rate by 11:00, there is still time to respond. If it is not identified until 15:00, recovery options are limited.

Project end-of-day backlog position hourly

At each hourly review, calculate: (opening backlog + emails received so far − emails cleared) + projected remaining arrivals − projected remaining capacity. If the projected end-of-day position exceeds the acceptable carryover limit, escalate immediately. Waiting until the end of the day to discover a miss is a process failure.

Agent redeployment between channels is slower for email

A voice agent can be moved from a quiet voice queue to a busy voice queue in real time. An agent moved from voice to email needs a context switch and typically produces lower throughput in the first 30–60 minutes. Build agent redeployment plans that account for this switching cost — if cross-channel blending is planned, it should be planned at the start of the shift, not as a crisis response at 16:00.

Do not treat email carryover as cost-free

Carryover backlog is not free. Each email that carries over adds to the next day's workload, increasing the risk of the next day also missing its clearing target. A small daily carryover compounds into a multi-day backlog crisis within 5–7 operating days. WFM must flag when carryover exceeds the agreed daily limit rather than normalising it as an acceptable outcome.

Email and ticket WFM questions

How do you calculate staffing requirements for an email queue?

Use a throughput model, not Erlang C. Formula: Agents required = (New volume + Opening backlog) ÷ (Agent throughput rate × Available hours). Example: 800 new emails + 120 backlog = 920 total. At 10 emails per agent-hour over a 10-hour day: 920 ÷ 10 = 92 agent-hours of productive time required. Divide by operating hours for agents needed: 92 ÷ 10 = 9.2, round up to 10. Apply shrinkage: 10 ÷ 0.75 = 13.3 → 14 agents on-shift. Validate at hourly intervals intraday using actual vs. forecast throughput and arrival rates.