PHAROS CONSULTING

WORKFORCE MANAGEMENT
101

FEB
2024
OVERVIEW
Workforce Management (WFM) is a critical call center function. It is the art and science of having just the right
number of staff, in the right place, at the right time. Staffing is a call center’s highest expense, so overstaffing can
easily result in overspending and missing financial targets. But understaffing can result in poor customer service,
overworked staff, and potential lost sales revenue. It’s a challenging balancing act.

WFM is an art because it’s about predicting the future and understanding human behaviour. Customers can choose
to call, email, text or chat at any time – not only when we want them to. Staff call handle times, unplanned leave
and attrition can increase for many different reasons. An accurate staffing plan will rely on the intuition, experience
in-depth knowledge of the business and its customer base. The best WFM plans will regularly use expertise and
opinions from across an organisation: especially from staff at the coal face.

WFM is also a science, because studying historical data and events, applying mathematical processes and using
specialist planning software can help find patterns and predictability in the randomness. Staff can then be
strategically scheduled to reduce the periods of long customer queues, or staff sitting idle. A Workforce Planner or a
dedicated WFM team will regularly make changes to call routing or staffing in real-time to balance out these quiet
and busy periods.

A modern challenge for all call centers is the “invisible” queue. Unlike a “visible” queue – for example, waiting at a
supermarket check-out or ATM – customers choose to enter a call center queue not knowing where they are in it, or
how long the wait will be. Most people expect to be answered straight away. As a result, frustration levels grow
higher as the delay grows longer. It’s incredibly important to provide a live answer before a staff member ends up
with a frustrated customer, or before the customer hangs up in the queue.

Staffing a call center is a challenge, and it isn’t easy. Even perfect plans can come undone due to random events.
Leadership, staff and WFM teams need to work as one to adapt when events don’t come to plan – but also celebrate
when things do.
WFM LIFE CYCLE
There are five key steps
involved in the WFM process. Data Collection & Analysis
WFM departments in large
organisations often have small - Historical call volume
sub-teams completing each of - Historical average handle time
these steps. - Historical staff utilisation

Real-time Management Forecasting

- Unplanned leave and off-phone - Time series data modelling
- Changing queue priorities to - Calls x AHT = workload forecast
maximise service levels - Workload forecast @ service
- Service level reporting level = staffing forecast

Scheduling
Staffing
- Staff ‘off phone’ time
- Overstaffing / understaffing cost
- Recruitment recommendations
and service level scenarios
- Staff rosters and pre-loading
- Budgeting
planned activities
DATA COLLECTION & ANALYSIS
The first step in the WFM cycle is to collect historical call and AHT data to base your forecast modelling from.

When considering the amount of data to use for forecasting, there are several considerations to make,
remembering that generally, more data is better):

• Ideally you will have at least 1 year of daily call volume and AHT data, to cater for seasonal changes in customer
call behaviour throughout the year (e.g. around Christmas, holidays, etc)
• You will also have at least 1 one month of intra-day call and AHT data, to cater for changes in customer
behaviour through the month (e.g. billing cycles) and throughout the day (calls volumes tend to be lower in the
morning and evening, peaking around the middle of the day)
• You want to have quality data – this is data that has no gaps or holes, and data not affected by business
structural changes (merging or separating queues or departments for example)
• Following the recent global pandemic, a thoughtful decision will need to be made about the data from 2020 –
2021 and how it impacted your organisation, and what your new normal looks like – are your patterns returning
to pre-pandemic levels, or are current patterns the new normal?
• Do you need to eliminate any of your data due to extraordinary past events? For example, if there was a natural
disaster that doubled your inbound call volume, if this excess is not removed, these calls will be included to
create your forecast.
FORECASTING
Once the historical data is ready, the next step is to use it to predict how much work a call center is likely to receive
over a given future period. Forecasting when work will arrive will allow us to place staff in the best possible position
to reduce wait times and increase service levels.
By using trusted statistical methods, planning software and subject mater expertise from the business, workforce
planners can estimate how many customer calls, emails, chat or SMS messages a center is likely to receive across a
given year, and right down to a 15-minute time interval.

Multiplying forecast volume against the forecast average handle time

(AHT, in seconds) - then dividing the outcome by the number of seconds
that 1 FTE can work - will generate a workload forecast: how many FTE
of work are expected to arrive for a given interval.
But unlike assembly line work (for example), call center workload peaks
and troughs depending on when a customer chooses to call. To ensure
service expectations are met, workforce planners input a target customer
service level (e.g. answering 80% of calls within 30 seconds) and the
workload forecast into a statistical formula to create a staffing forecast.
One of the most commonly used formulas is Erlang C:

The formula tests: ‘for a given workload and a given number of FTE, and
considering that work can arrive randomly, what is the chance of a
customer calling and having to wait outside the target service level.’
This step inevitably adds FTE to the workload forecast, but without it,
peaks and troughs are likely to create customer queues, resulting in poor
customer service, overworked staff, and lost sales revenue.
STAFFING
Getting to the “ideal” number of staff in a call center involves considering many trade-offs. It’s important to
understand the positive and negative impacts to staff, customers and costs as staff are added or removed.
It’s obvious that adding staff will improve customer service levels, and reducing staff will worsen them. What’s
interesting, however, is the degree to which service improves and declines with each staffing change.

Notice what happens to service, as staff are added or removed from a call center with a forecast staff requirement of
24 FTE:

If we require 24 staff, but add one more, service levels increase by 6% (from 85% to 91%) and staff occupancy - the %
of time staff are busy with customers, versus waiting for calls - decreases by 3% (from 83% to 80%). Adding a further
person gives less improvement, and the returns diminish with every additional staff member.
But when staff are removed, the impact is the opposite, and more severe. Reducing to 23 FTE reduces service levels
by 10%, and occupancy increases by 4%. If we are only 4 FTE short of our required 24 FTE, service levels are 0%,
customer queues are eight minutes long, and staff are 100% occupied (there is never a break between calls).
At this point, the entire system is inefficient and costly as customers are upset, tend to talk longer to staff, are less
accepting to sales and may hang up in the queue (calling back later, increasing telephony costs). If this continues for
weeks, staff begin to burn out and resignations increase. The high financial and reputational costs to understaffing
mean it’s better to plan, budget and recruit on the side of overstaffing (rather than understaffing) wherever possible.
SCHEDULING
After arriving at a ‘bodies in chairs’ staff requirement and evaluating staffing trade-offs, the next step is to create staff
schedules that will cover the workload as efficiently as possible. To determine a schedule requirement, the base staff
number is inflated to account for ‘shrinkage’ – the time that staff are being paid to work, but aren’t available to take
calls.
Staff aren’t available to take calls for 100% of the time that they are paid to
be working. Analysis into a call center’s historical shrinkage will show how Category Annual shrinkage
much additional planned and unplanned off phone time needs to be added, Annual leave 7.7%
on top of the time handling customer queries. Shrinkage tends to be Other paid leave 1.9%
somewhere between 20 and 40% (see example to the right). Paid breaks 5.9%
Going back to our example on the Staffing page (where 24 staff were Meetings 1.3%
required to handle call volume within a certain period), if we scheduled 24 Training 1.3%
staff, but 25% were unavailable to take calls, we would only have 24 x 0.75 = Coaching 1.3%
18 staff on phone. Looking at the table, 20 or less staff is expected give us a Customer Follow Up 3.6%
Unexplained 2.0%
service level of 0%. To add shrinkage, staff requirement is divided, not Total 25.0%
multiplied. Using the same example, 24 / 0.75 = 32 staff.

The next step is to create the actual staff work schedules. The secret of
successful scheduling is to have many different types of schedules as
possible. This enables the best fit from an efficiency standpoint, and
provides more options for call center staff. If a call center requires 32 FTE
for example, this might be made up with a large amount of part time staff,
and full time staff willing to work 4 x 10 hours per day. This provides a lot of
scheduling flexibility, if customer call patterns start to change over time.
Once schedules are in place and communicated, all planned off phone time
is loaded against them. Staff break times are then optimised to match
staffing coverage to the forecast staffing requirement as much as possible.
REAL-TIME MANAGEMENT
Once workforce schedules have been created and provided to staff, the day in question arrives. A workforce
planner’s next step to then manage those schedules in the moment, and making adjustments and re-route work to
ensure service levels are met each day. There are three key steps in the real-time management process:
• Tracking – service level in current interval and day to date, calls vs. forecast, AHT vs forecast, off phone vs
planned, staff occupancy. If service issues are apparent, the next step is to figure out what is causing the
problem.
• Communicating – the net staffing (+ or -) and communicate this to appropriate individuals and groups. In
extreme situations, customers may need to be contacted (e.g. putting messaging on IVRs)
• Reacting – find ways of getting more staff on the phones, getting staff out of planned off phone, technology
routing, cancelling certain activities (e.g. selling), O/T.

This doesn’t apply to just understaffing. Periods of overstaffing should be utilised, e.g. moving team meetings,
additional training to make the best use of business time and resources.

Data and insights from the day are collated, and form the basis of the next Data Collection & Insights phase. The
WFM life-cycle is complete, and continues in a loop of (hopefully) continuous learning, improvement and efficiency.
Reference Material
• Workforce Management Essentials: Principles and Programs for Contact center Staffing, Penny Reynolds,
[2015]

• Team WFM: Building a Collaborative Process for Contact center Staffing, Maggie Klenke [2018]