staffing and budgeting in an age of continuous...
TRANSCRIPT
STAFFING AND BUDGETING IN AN AGE OF
CONTINUOUS PLANNING
Ric Kosiba
Your Seminar Leader
Ric Kosiba is vice president and founder of the Interaction Decisions Group at Interactive Intelligence. He founded the company, Bay Bridge, in 2000 that was acquired by Interactive Intelligence in August of 2012.
Ric has been involved in the call center industry 20+ years and enjoys working with contact center analysts and helping companies maximize profitability through math modeling. He holds a B.S.C.E., M.S.C.E., and Ph.D. in Operations Research and Engineering from Purdue University (go Boilers!).
Ric is responsible for the development and enhancement of our contact center capacity planning and analysis product line. He thoroughly enjoys working with our brilliant development and operations research team, which helped Interaction DecisionsTM become the leading U.S. supplier of long-term forecasting and planning solutions.
Ric resides in Maryland with his wife and four children. He loves being a dad and enjoys coaching his kid’s football, basketball and lacrosse.
Today’s discussion
• How do you plan in this ever-changing environment?
• Specifically, we will touch on:
– Sources of forecast error
– Determining whether your forecast error is significant—or whether
something is changing in your environment
– Simulating your contact center so you can evaluate different planning
scenarios
– A process that allows smart analysts to improve efficiency while hitting
service goals
13,500,000 hits for “changing business environment”
History is often not a great guide!
• Been hearing (for the last ten years)
that contact volumes and other
important metrics are just not
forecastable.
• Forecasters are measured for their
“error” and are finding it tougher
every year.(blind models leading blind models)
What is forecast error?
• The general accepted definition of forecast error is variance between what was forecasted and what actually happened
• Error seems to be a strong word here
• What if we are forecasting in a time of change?
(now this is an error)
How do you know if there is “forecast error”?(hint: variance analysis!)
Should check all performance driver variance: volumes, AHT, attrition, shrink, outbound contact rates,…
Variance analyses are our canary in the coalmine
Why is there variance?
1. The forecast may be off– a math or process error
2. Something external may have changed the mix of contacts or agent performance
3. Something internal may have changed the mix of contacts or agent performance
4. It could be a random occurrence (a blip)
A very appropriate role of your forecasting process is to be an early warning signal
Is it error or is it change?
Error
• Can I tweak my model to make it
better?
• Hindsight: Was it my modeling
choices that made the difference
in accuracy?
• Nobody understands what’s
different this month
Change
• Or do I have to throw away my model for something radically different?
• Hindsight: Did I have any way of forecasting this right?
• We are hearing about the change (from the agents, from the news, etc…)
What most management teams view as error is really change! The best management teams view error as change!
Is the service impact of a forecast change significant?
Sensitivity analysis: A great way to measure the impact of a performance driver
Does the cost of the variance look significant?
Every change to a plan has a corresponding cost (or benefit)
A quick aside… service impact of shrinkage variance
Getting shrink correct is as important as getting volumes right!
What do you do when there is variance?
Significant Variance
Controllable?
Yes
Yes
Yes
No
NoNo
Want to fix?
Fix Issue!
Ready to makeA decision?
Prepare for change!
Monitor Performance
Several small “errors” can lead to real problems
• If their combination are crippling: for example, getting more calls
that have much higher handle times
• So we need a way to evaluate this!
Simulating your contact center so you can
evaluate different planning scenarios
I need a way to draw this curve!
Discrete-event simulation
Is the quarterback being rushed?
Is the receiver covered?
Is the quarterback running?
Is it a long pass?
Simulation is used because it is accurate
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
SER
VIC
E LE
VEL
(%
)
DAY #
Preferred Service: Service Level Comparison Simulation vs. Erlang-C vs. Actuals (Weekly Summary)
SL Actual SL Sim SL Erlang
Call Type
Simulation Service Level Prediction Erlang-C Service Level Prediction
Avg. Err Avg. Abs. Err
Std. Dev. Abs.
ErrAvg. Err Avg. Abs. Err
Std.
Dev.
Abs.
Err
Loans 0.59% 0.78% 0.65% 22.42% 22.42% 4.61%
Member Services 0.24% 1.19% 1.20% 25.97% 25.97% 5.56%
Preferred Services -1.27% 2.21% 1.75% 24.11% 24.11% 4.67%
Retail 0.86% 2.66% 1.39% 3.36% 4.03% 4.85%
Credit Card 0.31% 1.01% 1.08% 9.99% 9.99% 3.41%
Auto Insurance 1.20% 2.45% 2.27% -3.46% 3.46% 1.59%
Average 0.32% 1.72% 1.39% 13.73% 15.00% 4.12%
Tip: Validation of your analytic process breeds confidence in both your analyses, and you! Make validation a regular part of your planning meetings– even if everyone is tired reading how smart you are!
Service: ASA, SL, Abandon, Occupancy
Evaluating risk requires a (validated) simulation
With simulation, you can change anything and see resulting service (and vice versa). Accurately.
Capacity planning optimization algorithms
Mathematical technologies, such as integer programming will ensure that plans include just-in-time hiring and an optimal balance between hiring and overtime. These technologies will determine hiring solutions that determine the most efficient centers, understanding the differences associated with each center’s seasonal performance, handle times, attrition, sick time, etc…
Tip: Invest in automating and optimizing the creation of hiring /extra-time /undertime/ controllable shrinkage plans
A process that allows smart analysts to improve
efficiency while hitting service goals
SetGoals
Forecast Plan
Drivers
Build Operating
Plan
Perform Risk
Analysis
Create and Publish Budgets
Compare Performance to Plan and Budget
If I could automate and optimize this process, we could do some cool what-ifs!
A process that allows smart analysts to improve
efficiency while hitting service goals
SetGoals
Forecast Plan
Drivers
Build Operating
Plan
Perform Risk
Analysis
Create and Publish Budgets
Compare Performance to Plan and Budget
Forecasting Models
Discrete-Event Simulation Models
Optimization Models
Variable Labor Models
Variance Analysis
Optimization Models
How can you provide great analyses?
You need a capacity planning process that
is both quick and accurate!
Automate, optimize, and validate!
Ensure the operation is nimble and efficient in reacting to day-of changes within the environment
Ensure the right number of agents are trained and available to meet the seasonal and changing demand.
WFM System Strategic Planning
Scheduling Shift Bids Adherence/Conformance Real-time Management Etc.
Hiring Plans Budgets Establishing Optimal Goals What-if Analysis Etc.
Ensure that the right numbers of agents are available at the right time to service our customers.
Workforce Optimization
24
The Standard Planning Technology: The Erlang Spreadsheet
Erlang overstaffs 5-15%
Tedious and error prone. Turn around
time is days or weeks. Confidence in the
process is low.
Missing important metrics (e.g.
Abandons, Service achieved)
Single site models allocate staff
to multiple sites inaccurately,
and misses opportunities for
efficiencies
The relationship between staff,
calls offered, handle times and
service (including abandons) is
not directly addressed. The
concept of a “requirement” is
less accurate for strategic
planning
These models work in only
one direction: service to
requirements. They do not
work staff to service
expected, which is required
for what-if analysis
Determining when to hire and
when to offer overtime, is a
complete judgment call on the
part of the analyst. A bad day
can cost millions of dollars
(Ditch it!)
Forecasting do’s and don’ts
• Do update the underlying data consistently
• Do provide risk analyses when forecasts are in flux
• Don’t beat up your forecasters (they are people too- and times
may be a-changing)
• Do investigate why forecasts are off– internal, external, blip?
• Don’t just forecast volumes. Forecast all important metrics (like
shrink!)
Planning is important!
By automating planning:
• Consistent customer experience: less service variability
• No over-staffing bias and just-in-time staffing: lower costs
• Unparalleled executive-level contact center analyses and better understanding of the trade-offs associated with resource decisions
By optimizing the capacity plan, customers typically see a reduction in agent paid hours (5-10%)
What is Interaction Decisions?
• Decisions is a long-term contact center strategic planning and what-if analysis system. It helps to:
• Because it is fast and accurate:– Perform risk and sensitivity analysis of your contact center
– Evaluate center what-ifs: investments, consolidation, and growth opportunities
• Decisions complements traditional workforce management software by focusing on strategic decision making and long-term planning
ForecastRequirements
SimulationStaff & Capacity
Plan OptimizationBudget
Contact Us!
Ric Kosiba
410-224-9883
… if you would like a copy of the slides or to see a quick Decisionsdemonstration
Also! We have white papers all about planning and analysis at www.inin.com
(look for Strategic Planning)