RMT Engineering Logo
Contact centre workforce planning

Workforce Management

Erlang-C at 15-minute intervals, sized for the contacts that still reach a person once your AI agents have taken their share. Rosters build themselves against contracts and break rules. Adherence you watch while the shift is still running. Not the Monday after.

  • Forecast adjusted for measured AI containment
  • Reoptimised intraday
  • Adherence alerts during the shift
We reply within one business day. No newsletter.
Erlang-C Forecasting model
Intraday Reoptimisation as the day moves
Included No separate WFM licence
15-min Per channel and skill

Know how many people Tuesday at 14:15 actually needs

Traditional WFM tools forecast every contact as if a human will answer it. That assumption used to hold. In a centre where AI agents answer first, it overstaffs the floor by exactly the share the AI contained, and the planner ends up padding a number nobody in the room trusts. OptiML measures containment interval by interval and feeds it into the same model that builds the roster.

Staff to the forecast, not to the guess

Most planning teams work from a spreadsheet that was accurate on the Friday it was built. By Tuesday the queue mix has moved, two people are off sick, and nobody knows which interval is short until the service level has already gone. Then someone builds a report explaining it. Here, forecasting and scheduling and intraday correction all run off one set of intervals. The roster gets fixed during the day.

By the numbers

Few Facts - at a glance

  • Erlang-C

    forecasting model

  • Intraday

    reoptimisation

  • Included

    in the platform

The process

How It Works

The four stages, end to end

Four stages. Arriving contacts at one end, a published roster at the other, and an accuracy figure feeding back into the model. Each stage hands its output to the next, so the figure you review on Friday is measured against the forecast that built Monday. Nothing is exported to a separate planning tool and re-imported.

  1. Forecast the intervals

    Erlang-C runs over 15-minute intervals, dimensioned by channel, skill and queue, with weekday, month-end and holiday seasonality carried in the model. Containment is measured, not assumed. The volume the forecast sizes for is the volume still reaching a person.

  2. Build and publish the roster

    Schedules generate against the forecast, agent skills and availability, contract terms and break rules, with breaks placed where they cost the least service level. You review the week. Then you publish it to the agents.

  3. Reoptimise during the day

    When volume runs ahead of forecast on a queue, the system proposes the smallest correction that still holds the service level. Two breaks moved later. One floating agent pulled out of the retail pool. You see the diff before anything changes.

  4. Watch adherence, then feed it back

    Adherence is tracked live. Deviations alert the supervisor mid-shift, while there is still shift left to do something about. Afterwards the historical reports by agent, team and interval sit next to forecast accuracy per week and per channel — the number that tells you where the model still needs work, and the one most planning teams never open.

Features

Every capability you need in one module

1. A forecast that knows what your AI already handled

Erlang-C volume and staffing forecasts by interval, channel and skill, adjusted for what your AI agents are containing. That adjustment is the whole point. A WFM tool that assumes every arriving contact reaches a human keeps sizing the roster for work the AI already finished, and the error grows every time containment improves. Accuracy is tracked week by week and channel by channel, which is how you find out the model is wrong before the service level does.

  • Erlang-C by interval, channel and skill
  • Containment measured from what the AI agents actually resolved, then fed into the staffing model
  • Weekday, month-end and holiday seasonality
  • Accuracy tracked against what actually arrived

See the reporting behind it

Illustrative interface. Accuracy percentages and intervals are sample data, not a performance claim.

2. Schedules that build themselves

Automatic schedule generation against forecast, skills, availability, contract terms and break rules. Full shifts, part shifts and days off come out as a week you can review before anyone else sees it. Intraday reoptimisation then adjusts as the day diverges from plan, and break optimisation puts breaks where they cost the least service level. Every proposed change is shown as a diff first. Nothing moves on its own.

  • Generated against forecast, skills, availability, contracts and break rules
  • Full shifts, part shifts, days off
  • Intraday reoptimisation when the day diverges from plan, proposed rather than applied
  • A diff before anything moves

See the full CX platform

Illustrative interface. Names, shift times and service levels are sample data, not a performance claim.

3. Adherence you see during the shift

Schedule adherence tracked live, with deviation alerts to supervisors while the shift is still running. The break that overran by six minutes. The fourteen minutes unavailable with no code against them. Historical reporting by agent, team and interval is there for the coaching conversation afterwards. That conversation goes rather differently when the deviation was caught the same day rather than a fortnight later.

  • Live adherence, not a report the following Monday
  • Deviation alerts to the supervisor while there is still shift left to fix
  • Reporting by agent, team and interval

See the supervisor floor

Bring last month's volume

We will run the forecast against your own interval data and show you the roster it produces.

Book a session with our team

Illustrative interface. Names and adherence percentages are sample data, not a performance claim.

4. Shifts your agents can manage themselves

Shift bidding with rules you configure. A swap marketplace that checks eligibility automatically against skills, against contract and weekly hour caps and against minimum rest gaps, auto-approving where the rules pass and routing the rest to a team lead with the reason attached. Time-off requests follow the same path. The point is fewer scheduling emails, and fewer swaps that turn out on Thursday to have broken somebody's rest gap.

  • Shift bidding with rules you configure
  • Eligibility checked before the swap, not after
  • Blocked automatically when a swap would breach weekly hours or the minimum rest gap between shifts
  • Time-off requests through the same approval path

See the desk agents work in

Illustrative interface. Names, shifts and eligibility outcomes are sample data, not a performance claim.

5. Three programmes, one floor

Define skills and assign proficiency levels, then route on exact match first, partial match second and longest idle after that. Agent pools group capacity across teams and programmes. That is how a BPO staffs three client programmes off one floor without running three schedules and paying three planners. Skill is not a yes/no flag. Most WFM tools treat it as one, which is why a roster can look perfectly correct while the calls keep landing on the wrong person.

  • Proficiency levels, not a yes/no flag
  • Exact and partial matching, ordered by priority, then longest idle
  • One pool spanning several client programmes
  • Overflow capacity identified in advance rather than improvised at half past nine on a Monday

Talk to us about multi-programme rosters

Illustrative interface. Names, skills, levels and programmes are sample data, not a performance claim.

6. Attrition signals while they are still signals

Gamification with badges, leaderboards and streaks on real outcomes rather than on login time. Peer recognition and a kudos wall. Wellbeing pulse surveys with trend tracking, so a workload score that falls two weeks running shows up as a trend rather than as a resignation letter. Attrition risk is flagged to team leads while it is still a signal.

  • Badges, leaderboards and streaks on real outcomes
  • A kudos wall, and peer recognition that reaches further than one team lead inbox
  • Wellbeing pulse surveys, trended
  • Attrition risk flagged to team leads early enough to matter

See how coaching plugs in

Illustrative interface. Names, scores, streaks and survey results are sample data, not a performance claim.

Use Cases

Where Workforce Management delivers value

Containment-adjusted Erlang-C forecasting

The AI answered first, and the roster knew

A telecom customer service centre (illustrative)

Scenario

Voice agents now contain a large share of billing enquiries, but the planning spreadsheet still sizes every arriving contact as a human contact. So the 13:00 interval gets staffed for volume that never reaches a person. Rebuild the same forecast on measured containment instead, and the required headcount for that interval drops to what the queue actually needs.

Outcome

Planners stop padding the roster to cover a number nobody trusts. The accuracy report shows which channel the model still reads badly, rather than averaging the error away across all of them.

Intraday reoptimisation with a visible diff

One o'clock, and the billing queue is running hot

An online retailer (illustrative)

Scenario

Volume runs ahead of forecast on the billing queue mid-afternoon. Nobody wants a planner rebuilding the day by hand at 13:04. The system proposes moving two breaks later and pulling one floating agent out of the retail pool, then shows the change as a diff before anything is applied.

Outcome

The correction lands inside the hour it was needed. Break rules intact. The team lead can see which two people were affected, and why.

Skills, proficiency levels and shared pools

Three client programmes staffed from one floor

A mid-market BPO (illustrative)

Scenario

Billing, deliveries and collections belong to three different clients but share the same agents and the same room. Skills carry proficiency levels. One shared pool spans all three programmes, and routing takes exact match first, then partial, then overflow.

Outcome

One roster covers three contracts. A partial-match agent gets used as deliberate overflow capacity rather than as an accident nobody planned for.

At a glance

Specification

The numbers and limits, without the sales copy

Specification for Workforce Management
Specification Detail
Forecasting Erlang-C by interval, channel and skill, AI-containment adjusted
Intervals 15-minute intervals, dimensioned by channel, skill and queue
Seasonality Weekday, month-end and holiday patterns carried in the model
Scheduling Automatic generation, intraday reoptimisation, break optimisation
Adherence Real-time tracking with deviation alerts, historical reporting
Self-service Shift bidding, swap marketplace with eligibility checks, time-off requests
Eligibility rules Skill match, contract and weekly hour caps, minimum rest gap, auto-approval where rules pass
Skills Proficiency levels, exact and partial matching, priority ordering
Pools Cross-team and cross-programme capacity grouping
Engagement Badges leaderboards streaks peer recognition wellbeing pulses
FAQ

Questions,
answered

What teams ask us before they roll out Workforce Management — how it works, what it needs from your side, and what happens when it gets something wrong

Still not sure?

Talk to a specialist and get a straight answer.

Ask our team

OptiML Workforce Management needs your own history to build a first forecast: contact volumes at interval level and handling times by channel and skill, plus the agent skills, contract terms and break rules the roster has to respect. If that history sits in another WFM tool or in a spreadsheet, fine. Bring last month and we will run the Erlang-C model against your own intervals before you commit to anything.

Measured. Containment in OptiML Workforce Management comes from what your AI agents actually resolved in each interval, reported through the platform analytics, and it feeds the same forecast that sizes the roster. This matters more than it sounds. A WFM tool that assumes every contact reaches a human will keep staffing for work the AI already finished, and the gap widens every time your containment rate improves.

Intraday reoptimisation in OptiML Workforce Management proposes the smallest change that still holds service level, moving breaks or pulling a floating agent from another pool, and shows it as a diff for you to approve rather than applying it quietly. Separately, forecast accuracy is tracked per week and per channel. A channel the model consistently misreads stays visible instead of disappearing into an average.

Swaps and bids run through a marketplace that checks eligibility before the swap rather than after: skill match, the contract and weekly hour caps, and minimum rest between shifts. Where the rules pass, the swap auto-approves. Where they do not, OptiML Workforce Management blocks it or routes it to a team lead with the reason attached, so nobody has to work out on Thursday why Sunday is suddenly uncovered.

OptiML Workforce Management is part of the OptiML CX Platform, not a separately licensed WFM product bought from another vendor, so the roster, the containment number and the adherence report all come out of the same data. Deployment is hosted in the region you choose. Rosters, adherence records and forecast data stay inside that tenant.

Keep exploring

Additional supporting modules

Native modules that share the same record, knowledge and governance

Platform module

Supervisor Console

Where adherence and deviation alerts land. The live floor, two seconds old, with listen, whisper and takeover.

Explore
Platform module

Analytics & Reporting

76 report definitions across every channel, exportable six ways, including the containment number the forecast runs on.

Explore
Platform module

Agent Desk

Where the schedule meets the shift: presence states, capacity limits and the skills routing runs on.

Explore
Connected solutions

Where Workforce Management is used

The Solutions pages that lean on this module, and what it looks like once it is configured for a particular floor, job title or job to be done.

7 solutions built on this module

By industry

Sector floors that run on it

3

By role

Job titles that answer for it

1

By use case

Jobs it is put to work on

3
Browse every solution
Talk to a specialist

See it forecast your own volume

Bring last month's interval data and your break rules. We will run the Erlang-C model against them, show you the roster it produces and point at the intervals where your current plan is short.

  • 30 minutes
  • Your own interval data
  • No slide deck unless you want one

Book your slot

Leave your email and our team will come back to you within one business day.

or reach us directly
Call our team Email us

Your details stay private. We never share them.

This website uses cookies.

Cookies are small text files that allow us to create the best browsing experience for you on our site. By continuing to use this website or clicking "Accept & Close", you are agreeing to our use of cookies. To understand how we use cookies or how to manage them, please see our cookies policy.

Ask OptiML

Powered by RMT Engineering