Workforce forecasting

Workforce Forecasting

The refrain is common: we need to adhere to our targeted service levels… but at the same time, we also need to keep staffing costs down. Whether in a contact centre or bank branch environment, workforce managers everywhere face the constant challenge of balancing the priorities of service levels and labour costs. This is especially true in today’s economic environment where inaccurate forecasts can lead to poor scheduling decisions and can have dramatic impact on bottom line profitability. Workforce management (WFM) software has long been used to match demand in the form of calls, emails, web chats and other agent work with supply in the form of the agents themselves. Regardless of the sophistication of the WFM tool or processes, the foundation of optimising staffing levels lies in the precision of the forecasting engine.

More precise forecasting is a classic win-win scenario, enabling managers to more closely align demand and supply resulting in optimal cost and profit performance.

The laws of supply and demand

When demand (inbound calls, outbound calls, emails, web chats, etc.) is greater than supply (the agents themselves), the price, in the form of reduced service levels, falling customer satisfaction and poor agent morale, rises. When supply is greater than demand, service levels tend to improve, but at the cost of idle and unproductive agents. The key to optimising the bottom line performance of your contact centre is to find a harmonious balance between supply and demand. This bottom line performance is directly impacted by the direct costs of hiring and employing your agents, but it is also influenced by client satisfaction, agent morale and other factors.

The importance of precision forecasting

Let’s start by drawing a distinction between “forecast accuracy” and “precision forecasting”, as they are not interchangeable terms. Very simply, forecast accuracy can be defined as the difference between the amount of work per time period projected to be presented and the amount of work that is actually presented. In a real-world environment, this work may be inbound calls, outbound calls, emails, web chats, etc. In the interest of simplicity, we’ll just call these “calls”.

Many vendors advertise their forecast accuracy and many customers use this metric as part of their buying criteria when evaluating commercial WFM systems. However, forecast accuracy is only one component in the precision forecasting equation, albeit an important one. Another factor is how configurable the forecasting engine is to account for changing conditions.

By way of example, a linear forecast is one that says just as many calls will be presented today as were presented yesterday. This is as unrealistic as it is oversimplified. Many factors affect the demand presented to the average contact centre. The number of factors can be virtually unlimited, but includes regular events like: day of week (volumes may be higher on Mondays than they are on Wednesdays); day of month (volumes may be higher on the 1st, 15th or 30th); or season (holidays, anyone?).

For many contact centres, however, being able to create forecasts based on regular events by themselves is not enough. A catalogue retailer, for example, would need the flexibility to forecast the increase in calls soon after a catalogue drop. An entertainment ticketing company would need the ability to predict the demand that might be generated when tickets go on sale for a leading musical act. A public safety agency would need a forecast that accounts for an increase in calls as a result of a major convention or sporting event. You get the idea.

Campaign forecasting provides the ability to include those non-recurring or irregular events in the creation of a forecast. When coupled with the ability to account for the effects of regular events, the contact centre now has a very powerful, flexible tool to create very precise forecasts. Other factors that can affect the forecast include variations in the volumes of the types of calls offered, average handle time for each type, and abandon rates.

Here’s where the metric of forecast accuracy becomes meaningful. One way to look at forecast accuracy is that it provides a level of confidence as to how well the actual call volume will match the projections. Often, inaccurate forecasts are a product of not fully appreciating the impact of these regular or non-recurring events.

Forecasting Software

Before any staff planning can be done, an enterprise first needs to understand what has happened in the past.

A solution which provides historical data from entire customer contacts (i.e. from CTI as well as the ACD) means that scheduling can take place in a more realistic way. Enterprises should also be able to factor in exceptions, such as advertising campaigns, training and public holidays, and view when the best time for a meeting or training session will be, and measure the impact on the rest of the contact centre. Running regular hypothetical 'what-if' scenarios can show a scheduler how alterations to shift-patterns would impact performance, as well as assisting in business continuity by seeing what would happen in a flu epidemic, for example.

A great deal of unnecessary agent work can be removed by identifying the types of call that are being received, and determining whether these could be reduced further up the line, in the departments whose activities actively affect the volume and type of calls received, e.g. marketing or IT (for the website). As such, workforce management is increasingly being used as part of an overall quality or performance management suite, which can include quality monitoring, HR management and training as well as the traditional workforce management forecasts and schedules, as all of these factors affect each other.

For example, understanding when and how other departments will be operating means that workforce management tools can be used to forecast and schedule accordingly (e.g. about a new TV advert). Additionally, contact centre management is able to brief agents - via a desktop broadcast if at short-notice - about the correct responses and issues, as well as changing IVR prompts and messages to provide answers to the more simple answers, and managing agent skill-sets for relevant call groups.

Businesses should look for flexibility in forecasting functionality: situations can develop very quickly, so forecasts can become useless without the ability to alter schedules dynamically to reflect reality.

Completing the balance: scheduling the right agents

Now that a precise forecast has been developed, we shift our attention to the supply side of the equation – the agents. In an ideal world, there would always be enough appropriately skilled agents to handle each call as it arrived to satisfy service level, customer satisfaction, cost and revenue objectives. However, as we all know, the contact centre is rarely, if ever, a perfect world.

To state the obvious, every agent is different. They have different skills, competencies, proficiencies and schedule preferences. In addition, myriad external factors affect an agent’s ability to perform well. Completing the balance means creating schedules that match the right agents to the projected work based on these factors.

Relatively speaking, creating the initial schedule is fairly easy and most of the commercial WFM products do it pretty well. It’s when agents start calling in sick, or call volumes are different from what was forecast, or agents leave and return from meals and breaks early or late that the real fun begins. The best scheduling plan is not set in stone. Rather it can be easily adjusted by workforce analysts and supervisors who are close to the action.

Powerful process automation wizards can make this job much easier. Say an agent calls in sick or is going to be late for their shift. A couple of simple keystrokes in a wizard can note the agent’s disposition and immediately identify potential replacements based on user-defined factors. For those agents calling in sick, this wizard can also identify whether or not that agent had previously requested time off for that day that had been rejected. This is not only a powerful and efficient scheduling tool, it also helps identify agent coaching opportunities. Intraday management, the real-time shifting of agent schedules in the heat of battle, if you will, is also made much easier through process automation wizards and drag-and-drop user interfaces. Extending or shortening an agent’s schedule is easily accomplished in this manner. Advanced WFM systems will also provide immediate feedback regarding impact to service levels, workload fit (the relative balance between supply and demand) and agent costs.

Three top tips

Use recurring and non-recurring events to create a precision forecast. Forecasting best practices suggest that more accurate forecasts enable more effective scheduling.

Be prepared for frequent changes to your staffing plan. Agents call in sick, will be late, will need to leave early, won’t leave or return from meals and breaks on time for a variety of reasons. Streamline the schedule-change process to reduce administrative overhead (and associated costs!) and ensure key metrics continue to be met.

Consider your overstaffing and understaffing impacts independently. Have a keen understanding of what factors drive the business. If customer satisfaction is your top priority some overstaffing may be desirable, even if it results in increased agent costs. If revenue generation and profit maximisation are most important, then understaffing may be desirable.

Too much. Too little. Just right!

In most high-performing contact centres, the two key metrics most closely watched by management are targeted service levels and operating budget (commonly up to 90% in labour costs). Given differences in business priorities and resources, each organisation needs to analyse their unique situation to determine the right balance between these two competing needs.

When demand outpaces supply (understaffed condition), callers wait longer for their calls to be answered, lowering service levels. This also has the collateral effect of reducing customer satisfaction and potentially negatively impacting customer retention. In virtually every contact centre environment, customers represent the revenue side of the profit equation. At the risk of stating the obvious, lost customers means lost revenues. Another unintended consequence is poor agent morale leading to increased agent attrition.

When supply outpaces demand (overstaffed condition), agents are idle and non-productive. Service levels may be high, customers may be delighted, but money is essentially wasted due to these idle agents. Agent salaries represent the lion’s share of the variable cost component in a contact centre. Optimising these costs has a direct and positive impact on profits. This is especially true in today’s tightening economy. Like a see-saw, the operation of the average contact centre oscillates between these two conditions. With the precision forecasting and powerful, flexible scheduling capabilities of many advanced WFM solutions, these oscillations can be minimised (but not entirely eliminated), creating a profitable, harmonious balance between supply and demand.