How to Plan Support Capacity, From Forecast to Schedule

Here is how I plan support capacity, from forecast to schedule.

In short: Support capacity planning is the process of turning a ticket forecast into the number of agents you need, then into a schedule that puts them where tickets arrive. The core formula is required FTE = workload hours ÷ occupancy ÷ (1 − shrinkage) ÷ paid hours per FTE. A team forecasting 11,000 tickets a month at a 12-minute handle time needs about 24 agents.

You can check each input as you go. The shrinkage calculator builds your shrinkage percentage from its parts, and the Erlang calculator sizes live channels such as chat and phone.

This is the final part of The Support Operations Playbook. The previous part built a weekly support team scorecard. This part uses those numbers to plan how many people you need, and when.

What are the four stages of support capacity planning?

1. Forecast: how many tickets will arrive, by channel and week.
2. Workload: how many agent hours those tickets need.
3. Headcount: how many full-time equivalents (FTE) deliver those hours after breaks, meetings and time off.
4. Schedule: when those people work, so coverage follows demand.

A plan that stops at a headcount number leaves the hardest part, coverage, to chance.

How do I forecast support ticket volume?

Use at least 12 months of history so you can see seasonality, then adjust for three things:

• Customer growth. Tie volume to active accounts, not to the calendar.
• Known events. Launches, pricing changes and billing runs add contacts for a few weeks.
• Deflection and fixes. Count a reduction only once it shows up in the data, not when the project starts.

Say a team averaged 10,000 tickets a month last quarter. Active accounts should grow 6%, which gives 10,600. A release will add about 400. The forecast is 11,000 tickets. For longer horizons, repeat this for each month ahead and run each month through the steps below.

How do I calculate how many support agents I need?

Workload hours = forecast tickets × AHT (minutes) ÷ 60
Required FTE   = workload hours ÷ occupancy ÷ (1 − shrinkage) ÷ paid hours per FTE

forecast tickets   = tickets expected in the period (here, one month)
AHT                = average handle time per ticket, including after-ticket work
occupancy          = share of logged-in time spent handling tickets (0.85 = 85%)
shrinkage          = share of paid time lost to breaks, meetings, training and leave (0.30 = 30%)
paid hours per FTE = paid hours of one full-time agent in the period (about 160 a month)

Here is the worked example with the 11,000-ticket forecast:

Step	Calculation	Result
Workload	11,000 × 12 ÷ 60	2,200 hours
Occupancy at 85%	2,200 ÷ 0.85	2,588 hours
Shrinkage at 30%	2,588 ÷ 0.70	3,697 paid hours
Headcount	3,697 ÷ 160	23.1, rounded up to 24 FTE

Headcount math

From 11,000 tickets to 24 agents

Clarivoxx analysis

Forecast tickets11,000

Workload hours2,200

Paid hours needed3,697

Required FTE24

The formula

Required FTE = (tickets × AHT in hours) ÷ occupancy ÷ (1 − shrinkage) ÷ paid hours per FTE. Here 11,000 × 0.2 = 2,200 hours; 2,200 ÷ 0.85 = 2,588; 2,588 ÷ 0.70 = 3,697; 3,697 ÷ 160 = 23.1, rounded up to 24.

Every minute you take out of handle time lowers the first line, which is why reducing average handle time shows up here as fewer hours needed. To test your own inputs, run occupancy through the occupancy calculator and your leave, training and meeting hours through the shrinkage calculator.

When do I need Erlang C instead of this formula?

The formula above works for email and web tickets, because they can wait in a queue. Chat and phone are different: customers wait in real time, so you staff each half-hour interval against a target such as 80% answered in 20 seconds. That needs a queueing model. The Erlang calculator uses Erlang C, and how I calculate service level explains the target itself. Erlang C assumes nobody hangs up while waiting, so I treat its answer as an upper bound and compare it with real interval data.

How do I turn headcount into a schedule?

Headcount is a monthly number. Coverage is an hourly one. Pull arrivals by day and hour, then build shifts around that curve.

Arrival pattern

Monday carries 1.7 times Friday's volume

Clarivoxx analysis

0%12%24% of weekly volume

Monday
24%

Tuesday
21%

Wednesday
19%

Thursday
17%

Friday
14%

Weekend
5%

Monday needs about 1.7× Friday's coverage (24% ÷ 14%)Daily share = tickets arriving on that day ÷ tickets arriving in the week × 100

Stagger shifts and days off so coverage follows this curve instead of a flat 9-to-5 week.

Here Monday carries 24% of weekly volume and Friday 14%, so Monday needs about 1.7 times Friday's coverage.

• Stagger start times: early, core and late shifts instead of one 9-to-5.
• Move days off: some agents work Tuesday to Saturday so Monday is fully staffed.
• Cover time zones on purpose. Customers in other regions send tickets after your office closes.
• Put coaching, QA reviews and training in the quiet hours, so shrinkage costs the least.

How do I keep the capacity plan accurate?

Forecast accuracy = 1 − |actual tickets − forecast tickets| ÷ actual tickets

• Monthly: with 11,600 actual against 11,000 forecast, accuracy is 1 − 600 ÷ 11,600 = 94.8%. Watch the direction of misses too. Three months of under-forecasting is a driver problem, not bad luck.
• Weekly: check the scorecard. If SLA slips on the same days every week, the schedule shape is wrong, not the headcount.
• Daily: move breaks or pause project work when volume runs high.
• Quarterly: if hiring and ramp-up take three months, decide on a spring peak in winter.

Review cadence

Who keeps the capacity plan current

Clarivoxx analysis

Forecast misses by more than 5%Monthly accuracy review and updated driversSupport ops leadOwner assigned

SLA misses cluster on the same daysReshape shifts in the next weekly scheduleWorkforce plannerOwner assigned

Volume runs above plan todayMove breaks, pause project work or open overtimeTeam lead on shiftOwner assigned

Seasonal peak one quarter awayApprove hires a full recruiting and ramp time aheadHead of supportOwner assigned

What I've seen in practice

The first capacity plan I built left shrinkage at a flat 15% because it "looked about right." Once we counted training, coaching, team meetings and holidays properly, it was closer to 30%. That explained why we were always short even though the headcount math said we were fine.

On one team that served partners in the Middle East and French-speaking Europe, Sunday was a normal working day for many Gulf partners. Tickets piled up until Monday and SLA misses clustered there. Moving two agents to a Sunday-to-Thursday week fixed most of it without adding anyone. The headcount was right. The schedule was not.

The threshold I use now: if the forecast misses by more than 5% two months in a row, I rebuild the drivers before I touch headcount.

The capacity dashboard I review each month

Each month I compare the plan with what happened, using the inputs of the FTE formula.

Metric	Plan	This month	Last month	Change
Tickets	11,000	11,600	10,400	+1,200
Forecast accuracy	95%	94.8%	96.2%	−1.4 pts
AHT	12 min	12 min	12.5 min	−0.5 min
Occupancy	85%	85%	82%	+3 pts
Shrinkage	30%	30%	31%	−1 pt
FTE on staff	24	24	23	+1

Forecast accuracy is the row to read first. At 11,600 tickets the same math gives 24.4, so 25 FTE. If volume stays above plan for a second month, the forecast drivers need updating.

That closes The Support Operations Playbook. Each part fed this one: fewer repeat contacts, a smaller ticket backlog, better quality, shorter handle times, fewer handoffs and less volume all show up here as fewer hours needed. The customer support guide brings them together.

What to do this week

Calculate your real shrinkage for last month, line by line, and rerun the FTE math with it. If the result is higher than your current headcount, you now have the evidence for the staffing conversation.

Frequently asked questions

What is a good shrinkage percentage for a support team?

There is no number to copy. Shrinkage depends on your leave policy, training load and meeting habits. Mine landed near 30% once everything was counted. Measure yours from its parts rather than borrowing a figure.

What occupancy should I plan for?

I plan around 85% for email and ticket queues. A higher number looks efficient on paper, but it leaves no slack for spikes, and agents who are busy every minute burn out.

Is FTE the same as headcount?

No. FTE counts paid hours, headcount counts people. Two half-time agents are one FTE. Plan in FTE, then convert to people when you hire.

How far ahead should I plan support capacity?

At least as far ahead as it takes to recruit and train a new agent. If that is three months, your quarterly plan has to look six months out.

Sources

• The Erlang C Formula, Call Centre Helper: how Erlang C sizes live channels.
• What is call centre shrinkage and how to calculate it, Call Centre Helper: the shrinkage formula and what it includes.
• Erlang (unit), Wikipedia: the traffic unit behind the Erlang models.