A modern contact center routinely has terabytes of operational data available at any given time. “Big data.” When effectively analyzed, this information can help contact center management maintain consistent and appropriate service delivery across the seasonal peaks and valleys of contact volumes. Until recently, however, the methods used for such analysis and forecasting were inaccurate and often considerably off-base. Algorithms are now available that enable more accurate forecasting, evaluation, and optimization of operational strategies across seasons and years — introducing an intelligent new era of how contact centers are managed.
The importance of strategic planning algorithms
The contact center strategic plan or capacity plan focuses on resourcing the contact center network over the next week to 18 months. For a contact center executive, a capacity plan is the best big picture decision-making device available. This plan and the resourcing decisions it expresses is the overarching statement of how management wants to treat its customers and agents. Because a well-managed and funded strategic plan leads to a well-managed operation, it is a great aid to achieving wanted customer and agent satisfaction levels. With proper foresight, analyses, and algorithms, service failures that normally result from an unmanaged or inefficient plan can be avoided.
Advanced strategic planning systems have mathematical models that both simulate the operational performance under different planning scenarios and develop resourcing plans that are most efficient while still achieving service goals. When variance to the plan is noticed, these simulation and optimization algorithms are key to understanding the trade-offs between service, cost, customer experience, and revenues. These algorithms make plain the service, cost, and experience repercussions of alternative resource decisions and lead to better informed resourcing decisions.
Simulations are descriptive models; they describe how the operation will perform under different agent resource levels or customer contact volumes. Simulation models can be proved accurate through a validation exercise where the model’s predictions are compared to historical contact center performance through good service levels and bad. Once validated, descriptive models can be used as predictive models of future contact center performance. Proving model accuracy gives decision-makers confidence in the analyses that flow from these models. The best simulation models are multichannel (that is, simulates email, back office, inbound, outbound, chat centers), multi-skill, and multisite models. These models are also used to determine how many agents are needed week-over-week to ensure service delivery.
The best contact center resourcing algorithms are staffing optimization prescriptive models, which prescribe the best hiring, overtime, undertime, and controllable shrinkage plans that meet servicing objectives at least cost. These models ensure consistent service delivery as they achieve just-in-time staffing plans (as real-world constraints allow), never hiring too many or too few contact center agents.
The combination of predictive and prescriptive algorithms let analysts determine the optimal resource plan that will meet service goals at least cost under any expected scenario. This approach produces the “best” management decisions. Used by a clever analyst, these algorithms will also accurately predict the repercussions and risks of making the wrong resourcing decisions. Given that the future is unknown and variable, a creative analyst can quantify the operational risk of making the wrong staffing decision. These scenarios can be evaluated beforehand. For instance, performing the what-if analysis of what would happen to service if we staffed optimally for today’s forecast, but it was wildly off! This analysis could be used to alter the staffing decision and protect from this real-world possibility.
Strategic planning technologies
1. Data capture and storage. ACD, dialer, workforce management, payroll, and other systems’ data are appropriately mapped, summarized, and stored in the strategic planning database. This database serves to calibrate the forecasting and operational simulation models, to ensure the accuracy of the system. The database also works as a point of comparison for variance reporting (“actual” versus “planned”) and assists with general-purpose contact center reporting. It is the first, most basic component of any capacity planning system.
2. Forecasting. Time series metrics that are important to forecast will be stored, including contact volumes, sick time and other shrinkage metrics, probability of right party contact (outbound), handle times, agent attrition, and so on. This data should be forecasted at the appropriate level of detail (by contact center and staff group for center-specific metrics like sick time and attrition). The best systems offer multiple forecasting methods to choose from, like Holt-Winters or regression modeling, in order to ensure that the method fits the data being forecast.
3. Performance prediction and requirements generation. This component is an algorithm that predicts the relationship between the volume of contacts, the amount of staff, the handle time, and the operational performance expected (like service level, abandons, contacts handled, and agent occupancy). These models must handle multi-skill, multichannel, and multisite operations, and must be validated to be accurate under multiple planning scenarios. The most accurate algorithms are discrete-event simulation models, which consider customer patience and contact center efficiency when determining expected performance under varying scenarios.
4. Staff planning optimization. Simulation models develop week-over-week required staffing levels, but cannot develop an actual plan. Conversely, optimization models, like integer programming, can both automate and optimize this process. For contact center staff planning, an integer programming-based staffing optimizer will find the mathematically provable, just-in-time hiring and overtime plan that achieves service goals at least cost. This feature provides an efficient plan that still considers the real-world constraints and is a huge source of ROI.
5. Variance analysis and budgeting. While the other components serve to develop plans, variance analysis serves to score and monitor the plan’s execution. For example, the best companies note changes in customer demand, handle times, customer experience scores, and sales per contact by measuring and monitoring forecast accuracy. They view their forecasts as their operational baseline, and variance as operational change, and therefore executive decision-making points. By consistently measuring the operational variance to the strategic (or capacity) plan, executives can detect when unforeseen changes happen, and can react quickly.
When all of these components are brought together and automated, scenarios can be evaluated rapidly and with confidence, enabling a new class of contact center analytics.
Best planning practices
1. Avoid spreadsheets for building your capacity plans. It’s difficult to perform what-if analyses accurately in a spreadsheet. Analysts spend too much time and effort maintaining spreadsheets and all of their embedded formulas, and not enough time developing value-added analyses. New solutions are available that include the predictive, prescriptive, and descriptive models necessary to allow fast, accurate, and cost-saving analyses.
2. Demand model validation and the best modeling technologies. An unvalidated model is a guess at best. The most effective technologies for modeling multi-skill or multichannel contact centers are discrete-event simulation models. They can include contact routing, customer patience, inventory backlog (for back-office or email channels), abort rates (for outbound channels), and contact center efficiency. The best staffing models are integer programming based and include the ability to produce just-in-time hiring, overtime/undertime, and controllable shrinkage plans.
3. Actively manage shrinkage over the long term. Vacation, team meetings, leave, overtime, undertime, and so on are items that, if planned for in advance and managed, go a long way to ensuring the operation runs most efficiently. It is a common error, with customer service repercussions, to “flat-line” shrinkage, week over week, in the plan. The best organizations monitor and track the seasonality of shrinkage at the local contact center level.
4. Understand the risk of stretch goals. Many companies build into their capacity plans and budgets projects to be paid for with improvements in the main contact center cost drivers. This practice is dangerous. The more that stretch goals exist in a plan, the more the success of the budget is at significant risk. As projects are delayed, the budget and strategic plan must be adjusted. Project success now becomes the main risk of plan failure. It is better that stretch goals be managed separately from the operational plan.
5. Set up regular times to decide. Variance to plan, again, is a decision-making point. With the expressed purpose of making resourcing decisions upon a noticed variance to plan, meetings should be held regularly, typically monthly, with contact center managers, finance, and contact center forecasters and planners all in attendance. Establish a statement of the variance and possible options at the outset, and then present several possible scenarios for analysis, the purpose being to lay out to the executive decision-maker the options and the risks of each.
The age of the algorithm is here
As contact center operations become more complex, we need the ability to predict, manage, and control this complexity. When algorithms are automated, the room for error is reduced dramatically and questions can be answered in minutes rather than hours or days. The quality and speed of the plan improves, the time to decision is greatly reduced, and the quality of the decision itself is also greatly enhanced because many more scenarios can be considered. This new way of intelligent planning makes it all possible.
Visit us | www.inin.com
Strategic planning solutions. Contact center automation. Integrations to the most popular CRM applications. All the tools you need to differentiate the service experience you offer your customers. Learn how Interactive Intelligence can help. 