STAFFING AND BUDGETING IN AN AGE OF CONTINUOUS PLANNING. Ric Kosiba

Transcription

1 STAFFING AND BUDGETING IN AN AGE OF CONTINUOUS PLANNING Ric Kosiba

2 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 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 Decisions TM 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.

3 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

4 13,500,000 hits for changing business environment

5 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. (blind models leading blind models) Forecasters are measured for their error and are finding it tougher every year.

6 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)

7 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,

8 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

9 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!

10 Is the service impact of a forecast change significant? Sensitivity analysis: A great way to measure the impact of a performance driver

11 Does the cost of the variance look significant? Every change to a plan has a corresponding cost (or benefit)

12 A quick aside service impact of shrinkage variance Getting shrink correct is as important as getting volumes right!

13 What do you do when there is variance? Monitor Performance Significant Variance Controllable? No Ready to make A decision? No Yes Want to fix? No Yes Yes Fix Issue! Prepare for change!

14 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!

15 Simulating your contact center so you can evaluate different planning scenarios I need a way to draw this curve!

16 Discrete-event simulation Is the quarterback being rushed? Is the receiver covered? Is it a long pass? Is the quarterback running?

17 SERVICE LEVEL (%) Simulation is used because it is accurate % 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% Preferred Service: Service Level Comparison Simulation vs. Erlang- C vs. Actuals (Weekly Summary) 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! 30.00% 20.00% 10.00% 0.00% DAY # SL Actual SL Sim SL Erlang Call Type Avg. Err Simulation Service Level Prediction Avg. Abs. Err Std. Dev. Abs. Err Erlang-C Service Level Prediction Avg. Err Avg. Abs. Err Loans 0.59% 0.78% 0.65% 22.42% 22.42% 4.61% Std. Dev. Abs. Err 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%

18 Evaluating risk requires a (validated) simulation With simulation, you can change anything and see resulting service (and vice versa). Accurately. Service: ASA, SL, Abandon, Occupancy

19 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

20 A process that allows smart analysts to improve efficiency while hitting service goals Set Goals Compare Performance to Plan and Budget Forecast Plan Drivers If I could automate and optimize this process, we could do some cool what-ifs! Create and Publish Budgets Perform Risk Analysis Build Operating Plan

21 A process that allows smart analysts to improve efficiency while hitting service goals Optimization Models Set Goals Forecasting Models Variance Analysis Compare Performance to Plan and Budget Forecast Plan Drivers Optimization Models Variable Labor Models Create and Publish Budgets Perform Risk Analysis Build Operating Plan Discrete-Event Simulation Models

22 How can you provide great analyses? You need a capacity planning process that is both quick and accurate! Automate, optimize, and validate!

23 Workforce Optimization Ensure that the right numbers of agents are available at the right time to service our customers. WFM System Strategic Planning 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. Scheduling Shift Bids Adherence/Conformance Real-time Management Etc. Hiring Plans Budgets Establishing Optimal Goals What-if Analysis Etc.

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. Single site models allocate staff to multiple sites inaccurately, and misses opportunities for efficiencies Missing important metrics (e.g. Abandons, Service achieved) 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 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 (Ditch it!) 24

25 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!)

26 By automating planning: Planning is important! 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%)

27 What is Interaction Decisions? Decisions is a long-term contact center strategic planning and what-if analysis system. It helps to: Forecast Requirements Simulation Staff & Capacity Plan Optimization Budget 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

28 Contact Us! Ric Kosiba if you would like a copy of the slides or to see a quick Decisions demonstration Also! We have white papers all about planning and analysis at (look for Strategic Planning)