You’ll be short three people at 14:00 on Tuesday
know it today
Manual shift planning just copies last week. Workforce Intelligence analyzes your call flow, forecasts next week hour by hour, calculates the required headcount using Erlang C based on your targets, and flags gaps in the plan before you even notice them. It generates the draft; you make the publishing decision.
What each hour of the week brings
Forecasts are derived from your historical call flow, not from generic industry averages.
Day and hour profile
Historical inbound calls are aggregated by day of the week and hour. Monday at 10:00 and Thursday at 16:00 are analyzed separately; recent weeks carry more weight, while older weeks are quietly phased out.
Not a single number, but a band
For each hour, low, medium, and high scenarios (p10, p50, p90) are displayed side by side. Whether you staff for the average and monitor peaks, or staff for the peak, is your decision; the band doesn’t hide the choice.
Official holiday calendar integrated
Public and official holidays are read from the system calendar. A holiday is not treated like an ordinary Tuesday in forecasting; it is also skipped in the draft plan.
Empty hours are data too
No calls at 03:00 is not a missing record, but information. The profile also learns zero; staffing is not written for that hour in vain.
How many people are needed
You set the service level target; the system calculates how many people need to be on the line per hour to reach that target.
Calculation with Erlang C
The Erlang C formula, used in call centers for decades, provides the number of people required for the target service level based on hourly call volume and average handling time. The target is yours: for example, 80% of calls answered within 20 seconds.
Break and absenteeism allowance
The number of people required on the line is not the same as the number required in the staffing plan. If break, training, and absenteeism allowances can be measured, the measured value is used; if not, it is clearly labeled as an assumption and not presented as measured data.
Understaffed and overstaffed hours
The current plan is placed side by side with hourly needs. The table shows how many people are missing or excess in each hour; the worst hours are listed at the top.
Draft shift, no unapproved publishing
The system suggests a draft plan that covers understaffed hours, considering leave, official holidays, and weekly limits. The draft is not published automatically: the manager approves it with the existing button in the shift planner or deletes it in one click.
Screens: Weekly workload and staffing
Images captured from the real panel with sample data. Contains no customer data.
Click a step or keep scrolling; the screen changes as you go.
Weekly forecast
Expected calls by day and hour for the week; low, medium, and high bands together. Peak hours and your current mode are displayed at the top.
Understaffed and overstaffed hours
Current schedule alongside hourly needs. The table shows how many agents are missing or in excess at each hour, with the worst hours listed at the top.
Rationale and accuracy
CALLAII's rationale sentence built from calculated figures, alongside an accuracy card comparing last week's forecast with actual results.
The deeper your data, the further the calculation goes
We do not promise much with little data. The system checks the data it has and clearly states its operating mode on the screen.
- Volume mode: while data is still scarce, the weekly heat map, busiest hours and peak load are shown. Staffing calculation is not enabled at this stage; showing an unreliable number is worse than showing none.
- Staffing mode: once call volume and history are sufficient, the Erlang C calculation, understaffed and overstaffed hours, and draft shifts open automatically. Mode selection is automatic; the panel displays which mode you are in.
- Three weeks of incoming call records are sufficient for the initial profile. Since calls flowing through your PBX are already recorded, no separate data upload is required.
- The accuracy card compares last week's forecast with actual calls and displays the deviation (WAPE). If there is insufficient data for comparison, the card states "insufficient data" and does not fabricate an accuracy rate.
Reasoning alongside the number
Seeing a number on the screen is not enough; managers must be able to explain to their team why that number was generated.
CALLAII writes the reasoning
A brief reasoning sentence is generated for each calculation: why a specific hour is understaffed, why a specific day differs from last week. The sentence is constructed solely from calculated figures; CALLAII does not add its own estimates to the numbers.
Same screen family as the shift planner
Forecasting, staffing, and draft shifts reside within the existing shift planner and HR module. Approval, publishing, swapping, and leave workflows operate as is; no separate tool needs to be learned.
Your data stays yours
Each company's profile is built solely from its own calls. One company's data does not mix into another company's forecast.
On our own servers
Calculations and reasoning generation run on Callmenta's own servers; your call data is not sent to a third-party service for forecasting.
Generating numbers is easy; getting the right number is hard
Staffing calculations are based on queueing theory, forecasts include confidence bands, shrinkage is measured, and an accuracy score is provided weekly. All four steps are visible on screen.
Erlang C, the call center standard
The required number of agents on the line is calculated using queueing theory based on hourly call volume, average handle time, and target service level (e.g., 80% of calls answered within 20 seconds); it’s not just a rough division.
Backup plan with p90
The upper band is always displayed alongside the median forecast to show what happens if the staffing based on the average isn’t enough. Planning for peak volumes during campaign weeks is possible at a glance.
Measured shrinkage
Shrinkage for breaks, training, and absenteeism isn’t a fixed assumption; it’s measured from your break and attendance records. If data is insufficient, the value is labeled as an assumption, not presented as measured.
Accuracy card scores weekly
Last week’s forecast is compared with actual calls, and the deviation (WAPE) is displayed on screen. If the forecast isn’t better than simply copying last week, it says so; if there isn’t enough data, it states that too.
The difference between manual planning
Spreadsheets copy the past; Workforce Intelligence calculates the future.
- Instead of copying last week, the day and hour patterns of the last six weeks are weighted; recent weeks carry more weight, while older weeks are quietly phased out.
- Instead of a fixed shrinkage rate, a rate measured from your break and attendance records; if it can’t be measured, an honest assumption badge.
- Instead of a single number, low, medium, and high bands; the decision based on peak hours is yours, the system doesn’t hide options.
- The holiday calendar is not entered manually but read from the system; holidays are not treated like ordinary Tuesdays.
- Next to each figure, CALLAII provides the rationale: why a specific hour is open, why a given day differs from last week.
- No big claims with little data: the staffing calculation unlocks automatically once sufficient data is available; until then, it displays the workload heatmap and peak hours.
FAQ
How much data is required?
Approximately three weeks of incoming call records are sufficient for the initial volume profile. Staffing calculations automatically activate once sufficient volume and history are available; until that threshold is reached, the screen remains in volume mode and clearly states this.
Does it automatically publish shifts?
No. The system only generates drafts. Publishing is done via the existing button on the shift planning screen and requires manager approval; unwanted drafts can be deleted in one click.
Who sets the service level target?
You do. The target percentage and response time (e.g., 80% of calls answered within 20 seconds) are kept in company settings; staffing calculations are based on this target. When the target changes, requirements are recalculated.
Where do break and absence allowances come from?
If break and attendance records exist in the system, allowances are measured from them. If not, the default in company settings is used and labeled as an "assumption" on the screen. The two are never confused.
Which channels are covered?
Inbound voice calls, live support (website chat), WhatsApp, and outbound calls. Each channel is calculated in a separate tab; the combined view aggregates the required headcount per hour with channel breakdowns. For chat channels, the number of concurrent chats handled by a single agent is taken into account; since outbound calls have no queue, their requirement is derived from workload.
Try it with your own data
In the demo call, we map your call flow together: which hours are busy, how many agents you need based on your goals, and which hours are uncovered in your current plan.