WorkForce Forecasting Overview

1
Agenda

What we’re Going to Cover

The Forecasting Model

Forecasting the Business Activity

Determining the Labor Required to be Scheduled

2
The Forecasting Model

3
The Forecasting Model

Organizational
Structure

Labor Conversion Location

Methods Characteristics

Accurate
Forecasting

Budget Hours Historical Business

(optional) Actuals

Business Calendar

4
Forecasting Model Options
‘Top Down’ Model (simpler)

▪ Utilize existing Budget hours

▪ Step Standards
▪ Single Business Driver
▪ Single Task
▪ Simple Business Calendar
▪ Simple location characteristics

‘Bottom Up’ Model

(typically more sophisticated)

▪ Sales per Labor Hour

▪ Time Standards
▪ Multiple Business Drivers
▪ Multiple Tasks
▪ Hierarchical Driver data
▪ ‘Cross Driver’ Forecasting
▪ Derived Drivers
▪ ‘Time Shifting’
▪ Advanced Location Characteristics
5
Forecasting the Business Activity
What does your future business look like?...

6
Forecasting the Business Activity

Historic Business Unique Site Accurate Business

Actuals Data Analysis (Critical) Forecast

• Sales/Items/Transactions • Trends • Volume AND Shape

• Traffic • ‘Business Calendar’ • By week/day/15 mins
• Deliveries • Week and Day ‘Types’ • Configurable ‘Window’
• Prescriptions • ‘Moving’ Holidays • Multiple forecasting
• N years of Data • Seasonal Changes methods

7
Business Calendar

• Made up of Week and Day ‘Types’ (patterns and profiles)

• Exception Weeks (e.g. Christmas / Sales / School Holidays)
• Exception Days (e.g. Black Friday / New Year’s Eve)
• Typically 20+ ‘Normal’ Weeks…

• ‘Moving’ Days – Same date, different day

• Independence Day, Christmas Day etc.
• Analyse specific dates rather than just days

8
Forecasting the Business Activity
Week Types Day Types

Week Start Week Type Shape Date Day Type Shape (Profile)
Date (Pattern)
6/30/2013 Normal 1
5/5/2013 Normal A
7/1/2013 Normal 1
5/12/2013 Normal A
7/2/2013 Normal 1
5/19/2013 Busy B
7/3/2013 Normal 1
5/26/2013 Memorial Week C
7/4/2013 Independence Day 2
6/2/2013 Busy B
7/5/2013 Normal 1
6/9/2013 Normal A
7/6/2013 Normal 1
6/16/2013 Normal A
6/23/2013 Sales D
6/30/2013 Independence Week E
7/7/2013 Busy B
7/14/2013 Normal A
7/21/2013 Normal A

Weekly ‘Pattern’ A Daily ‘Profile’ 1

9
Forecasting the Business Activity
Example shape of business across a week in Grocery

Mon Tue Wed Thu Fri Sat Sun Week Total

% Split by
Day 13.84 13.65 14.57 14.02 12.85 13.96 17.12 100 %
Sales 26,133 25,787 27,511 26,471 24,273 26,364 32,331 188,870
10
Forecasting the Business Activity
Example shape of business across the day

10:30 10:45 11:00 11:15 11:30 11:45 12:00

% Split by
Time 1.42 % 1.61 % 2.23 % 1.59 % 1.53 % 1.82% 1.79 %
Sales ($) 469.88 531.63 736.46 525.98 507.76 603.09 590.38 11
Forecasting the Required Labor
Right role, right place, right time...

12
Forecasting the Required Labor

Accurate Business Labor Accurate Forecast of

Forecast Calculation Labor Required

Multiple ‘Conversion Methods’

• Shaping the Budget
• Productivity Measures (SPLH)
• Step Standards
• Volume AND Shape • Time Standards • By Task/Activity
• By week, day and time • Week/Day/15 mins
of day Location Characteristics
• Min/Max Staffing Levels
• ‘Fixed’ Tasks
• Unique Location Layouts
• Sq. Ft./No of Entrances
• Services/Departments
13
Labor Conversion Methods
Converting the forecasted business amount into required labor hours...

14
Conversion Methods
• A labor model converts the predicted business levels into an amount of work required by 15-minute
timeslots to service those levels

• It is used when scheduling to ensure the right people are in the right place at the right time

• Different conversion methods are available:

• Shape Existing Budget

• Driver (Sales) per Labor Hour

• Target Labor to Sales Ratio

• Step Standards

• Time Standards

• Fixed Staff Profiles

15
Shape Existing Budget

• The budget hours for the division will be spread across the week and days
according to the forecasted shape of business
Budget Hours = 250

• Hours will also be spread throughout the day based on daily profiles
• Minimum and maximum staffing levels are adhered to so it is possible the
budget will be broken based on these constraints

16
Driver (Sales) Per Labor Hour (SPLH)

• Divides the forecasted driver values in a timeslot by a pre-defined driver per

labor hour

• E.g. if using sales you could have a sales per labor hour (SPLH) of $100

• This means that every $100 of sales results in 1 hour of work

17
Target Labor to Sales Ratio (LSR)

• Multiplies the forecasted driver values by a target labor

to sales ratio % (LSR)

• E.g. if using an average pay rate of $20 per hour and target LSR
of 5% then $100 of sales would result in 0.25 hours

$100 sales X 5% LSR = $5 of labor

@ $20 per hour = 0.25 hours

18
Step Standards
• Calculates the number of employees required based on $
the total forecasted driver value
300
• Applies the same headcount for the entire shift (or time 200
band) but multiple time bands can exist per day 100
0
• There are 2 types:
1. Linear – Gap between steps is consistent e.g. $
Every $100 = 1 employee
300
2. Tabular – Gap between steps is inconsistent
200
$ Head Count
0 – 200 100
1
0
201 – 300 2
301 – 600 3 19
Time Standards

Multiplies the business driver value for each timeslot by a time standard
For Example:
It takes on average 2 minutes to sell the
snacks to each customer at a theater.
Based on previous historical data we are
expecting 6 people will buy snacks during
the 11:00 timeslot.

Therefore to service the 6 customers

this will take 12 minutes of work

20
Fixed Staff Profiles
• Also referred to as minimum/maximum staffing

• Can be fixed to specific times during the day or anchored against

opening/closing times

• Can be set differently for each day of the week

Minimum Staffing
3 Day From Time To Time Min Max
Head Head
Count Count

Every Day
30 minutes
2 2 2 2 2 2 2 2 2 2 2 Opening 1 10
before opening

Every Day Opening Closing 2 99

Every Day Closing 30 minutes after 1 10

1 1 1 1 1 closing

<- Opening and closing times ->

0 0 0 0 0
8:00 8:15 8:30 8:45 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 19:15 19:30 20:00 21
Labor Conversion Factors
For the more sophisticated model...

22
Fine Tuning Demand

• The labor demand calculation using the conversion methods we have just looked at
will be mathematically accurate however may not be very practical for scheduling
against.

• A number of fine-tuning options are available to improve the practicality of the

calculated demand:

• Rounding

• Smoothing

• Time Shifting

23
Rounding

• Raw demand is calculated as a number of minutes

15

• This needs to be rounded to 15 minute intervals to

provide a whole number of employees required for a

Mins
timeslot 7.5

• For example if 8 minutes of demand is calculated in a 15

minute timeslot, should this be 0 or 1 people required?
Work
• The rounding threshold can be configured
50 % Rounding
• ‘Slack time’ can be created when rounding up Threshold

Slack Time

Rounded Down so no
work 24
Smoothing

• Smoothing will spread spiked values across a configurable number of

adjacent 15 minute timeslots (typically 2 on each side)
• Removes peaks and valleys
• Provides a more practical shape to schedule against

80

20 25 30 40 30
25 20 27 27

25
Time Shifting

• Not all work takes place at the timestamp the data was received.

• Time Shifting allows you to define how much of the work should be moved either
forward or backwards in time, and by how far in either direction that work takes place.

• For example a cell phone salesperson may help a customer with a sale starting 30
minutes before the transaction is recorded in the point of sale system:

In this example the work is spread over

4 timeslots before the transaction time
and split proportionally 20/20/30/30

26
29