WFM guideSmall operations

WFM for small contact centres

Q: Do small contact centres need WFM software?

Contact centres with fewer than 20–25 agents can typically manage with a spreadsheet-based approach to forecasting and scheduling, using a free Erlang C calculator (such as Turnella's) for headcount calculations. Between 25 and 50 agents, the scheduling complexity increases enough that a dedicated WFM platform starts to pay for itself — particularly if the operation has multiple shift patterns, skill groups, or significant volume variance. Above 50 agents, the cost of manual scheduling (analyst time, scheduling errors, suboptimal shift patterns) almost always exceeds the cost of a WFM platform licence. The break-even for WFM software investment is typically: (WFM platform annual cost) ÷ (scheduling efficiency saving per FTE per year). At 50 agents and a 3–5% scheduling efficiency improvement, the typical break-even is 12–18 months.

Q: How do you schedule a small contact centre without WFM software?

A practical approach for small contact centres (under 25–30 agents) without WFM software: (1) Calculate weekly volume by day-of-week from 4–8 weeks of historical ACD data. (2) Break each day into 30-minute intervals and calculate arrival rates per interval. (3) Use an Erlang C calculator (such as Turnella's free voice calculator) with the peak interval arrival rate, your target AHT, and your SL target to get the minimum agents required per interval. (4) Build a shift grid in Excel or Sheets that shows required agents per 30-minute interval and assign shifts to cover the intervals that need the most agents. (5) Calculate shrinkage using the simple formula: scheduled agents = required agents ÷ (1 − shrinkage%). Review and update the model monthly. The largest risk is using the average arrival rate across all intervals rather than the peak — this always understaffs the peak period.

Small contact centres face the same WFM maths as large ones — but they face it without the economies of scale, the dedicated WFM analyst, or the software investment that large operations take for granted. A 15-agent team running on intuition and spreadsheets is not doing WFM "informally" — it is absorbing the cost of poor planning silently through unnecessary overtime, chronic understaffing, and staff who leave because the schedule is unworkable.

Why small contact centres are harder to plan than large ones

Higher statistical variance

Erlang C assumes arrivals follow a Poisson distribution. At large volume (500 calls/hour), actual volume in any 30-minute interval is close to the forecast. At small volume (10 calls/hour), actual volume varies 40–70% above and below forecast in any given interval. Small operations must plan for much wider variance — meaning more buffer headcount relative to average staffing.

Planning implication

A 15-agent operation should plan for 15–20% more variance than a 150-agent operation on the same SL target. Under-buffering is the most common cause of SL failure in small contact centres.

No redundancy

In a 10-agent team, one unplanned absence removes 10% of capacity. In a 100-agent team, one absence is 1% of capacity. This is why a single absence can collapse service level in a small operation and barely register in a large one.

Planning implication

Small teams should have an on-call or standby arrangement — at minimum, a list of agents who can be called in at 2-hour notice. Without this, absence-driven SL failures are structurally inevitable.

All-rounder requirement

Large operations can specialise: voice agents, chat agents, email agents, each skill-group sized independently. In a 10–20 agent operation, every agent must handle every channel and every contact type. This simplifies scheduling but means agent development takes longer (more to learn) and AHT is higher than a specialist.

Planning implication

WFM planning for small operations must use a single blended AHT across all channels. Do not apply Erlang C separately to voice vs. chat — blend them.

Practical WFM without a platform (under 25–30 agents)

5-step weekly planning process

Collect 4–8 weeks of interval data

Download 30-minute interval call volume from your ACD (every telephony platform has this — Avaya CMS, RingCentral Analytics, Twilio, Zendesk Talk, etc.). Build a table: day × interval × contacts. Average across the 4–8 weeks to get a typical day profile. Identify your peak interval (usually Mon 09:30–10:00 or 10:00–10:30 for most UK operations).

Calculate AHT from the same data

Average handle time = (talk time + hold time + ACW) ÷ contacts handled. Use the ACD data for talk time and hold time; ACW may be in a separate WFM or CRM system. Use weekly average — do not use a single-day sample. If you cannot extract ACW from your system, add 10–15% to talk+hold as a proxy.

Run Erlang C for your peak interval

Input: contacts per 30 minutes at peak, AHT in seconds, SL target (% of contacts answered within X seconds). Use Turnella's free Erlang C calculator. The result is the minimum number of agents required to meet your SL target during the peak interval. This is your peak staffing requirement.

Apply shrinkage to get scheduled headcount

Scheduled agents = required agents ÷ (1 − shrinkage%). A typical small contact centre shrinkage is 25–30% (absence 6–8%, training 4–5%, breaks 10–12%, admin/meetings 4–6%). If Erlang C says you need 8 agents in the peak interval and your shrinkage is 27%, you need to schedule 8 ÷ 0.73 = 11 agents for that interval.

Build the shift grid

In Excel or Google Sheets, create a grid: rows = 30-minute intervals (08:00 to 20:00 if that is your operating window), columns = agents. For each interval, enter 1 if the agent is scheduled to be available. Sum the column to get agents available per interval. Compare to the Erlang C requirement per interval. Close gaps by adjusting shift start/end times. Review monthly.

When to invest in WFM software

Agent count	Recommended approach	Break-even on WFM platform
Under 15 agents	Spreadsheet + free Erlang C calculator. Owner or TL manages scheduling manually. 2–3 hours/week.	Not applicable — WFM platform cost typically exceeds efficiency saving below 15 agents.
15–30 agents	Structured spreadsheet model (5-step process above). Consider a lightweight scheduling tool (Sling, Deputy, Rotaready) for shift management. WFM platform is an option but rarely essential.	18–36 months on a basic WFM licence at £400–600/seat/year at 15–25 agents.
30–50 agents	Strong case for a WFM platform, particularly if multi-skilled, multi-channel, or variable hours. Scheduling complexity typically requires 4–8 hours/week of analyst time without a platform.	12–18 months. At 40 agents, a 3% scheduling efficiency improvement (~1.2 FTE) typically justifies the platform cost in year 1.
50–100 agents	WFM platform is standard. Without one, scheduling errors, adherence gaps, and forecast errors are costing more than the platform licence every year.	6–12 months. At 50+ agents, the ROI case is almost always positive in year 1 at standard industry efficiency improvement rates.
100+ agents	WFM platform is table stakes. Full WFM analyst role (1 per 50–75 agents) is required. Platform cost is typically less than 1% of the labour budget it manages.	Under 6 months. The operational risk of not having a platform at 100+ agents (real-time adherence alone justifies the investment) exceeds the licence cost.

Three most common mistakes in small-operation scheduling

Using average volume, not peak volume, as the Erlang C input

What goes wrong

If your peak interval has 15 contacts and your daily average interval has 8 contacts, using the average produces a staffing plan built for 8 contacts/interval. During every peak interval (typically 4–6 intervals each morning and early afternoon), service level collapses. This is the single most common cause of SL failure in small contact centres.

The fix

Always use peak-interval volume as the primary Erlang C input. Build a schedule that covers the peak. Off-peak periods will naturally be overstaffed — this is correct; it is the statistical buffer for variance.

Applying no shrinkage (scheduling required agents, not scheduled agents)

What goes wrong

If Erlang C says 8 agents are required and you schedule exactly 8 agents, every absence, break, or training event immediately takes you below the required level. At 25% shrinkage and 8 required agents, you need 11 scheduled agents to have 8 available. Scheduling 8 means you typically have 6 available — 25% understaffed against the Erlang C requirement.

The fix

Apply a shrinkage multiplier to every Erlang C output. For a small team, use 25–30% shrinkage (higher than large operations because relative absence impact is greater). Scheduled agents = required ÷ (1 − 0.27) = required × 1.37.

No on-call or standby provision

What goes wrong

In a 15-agent team, a single unexpected absence removes 6.7% of capacity. Two absences removes 13.3%. Without an on-call arrangement, one bad absence day produces SL failure that cannot be recovered until the next shift. Small operations typically experience 1–2 unexpected absences per week — this is not exceptional; it is normal.

The fix

Maintain a list of 2–3 agents who have agreed (with appropriate notice and contractual terms) to be available for on-call cover on their scheduled day off. This does not need to be every agent — a rotating on-call rota covering 2 agents per week provides the resilience a small team cannot achieve through pure scheduling.

Small CC WFM questions

Do small contact centres need WFM software?

Under 20–25 agents: a spreadsheet-based approach with a free Erlang C calculator (like Turnella's voice calculator) is sufficient. 25–50 agents: a WFM platform starts to pay for itself — break-even is typically 12–18 months. Above 50 agents: WFM software is standard and the scheduling complexity almost always justifies the investment.

How do you schedule a small contact centre without WFM software?

(1) Collect 4–8 weeks of 30-minute interval volume from your ACD. (2) Calculate average AHT. (3) Run Erlang C on your peak interval volume. (4) Apply shrinkage: scheduled agents = required ÷ (1 − shrinkage%). (5) Build a shift grid in Excel and compare staffed agents per interval to the Erlang C requirement. The largest planning mistake is using average interval volume instead of peak interval volume.