Skills-based routing (SBR): get the gain without the pain!

Skill-based routing (SBR) has been around almost since the dawn of call centers. Since then, call centers have evolved into omni-channel contact centers, handling not just calls but emails, web chats, social media interactions, and so on. This means that SBR is more relevant than ever.

What is skills-based routing?

Skills-based routing (SBR) simultaneously queues calls and other customer interactions to more than one group of agents. Depending on the center's needs, the administrator can define rules that distribute incoming interactions in a preferred order. SBR routes interactions to the first available agent with the required skillset. SBR enables contact centers to derive the maximum value from a multi-skilled workforce.

Why skills-based routing is ‘A Good Thing’

• Customers are more likely to have their issues resolved on first contact. That is because they are routed to the agents who are most qualified to handle their issue.
• Agent occupancy and productivity are boosted. Without a multi-skilled workforce and SBR, separate teams of agents are dedicated to a single type of contact, differentiated by subject or contact channel, for example. SBR decreases agent idle time between customer interactions because agents with more than one skill will naturally spend less time waiting for an interaction they are qualified to handle. The occupancy and utilization of multi-skilled agents will therefore tend to be higher than that of single-skilled agents. The greater the extent of multi-skilling, the fewer agents are required to handle the workload. This effect is known as pooling efficiency.
• SBR reduces the need to fire-fight by moving agents from one type of interaction to another to follow differing peaks in workload. The system automatically finds the best resource for the workload mix, in real-time.

Having teams of agents dedicated to a single type of interaction offers no pooling efficiency. Consequently, in some centers, the goal is to train every employee to be a ‘universal agent’ who can handle any interaction. This will deliver the maximum possible pooling effect and the greatest schedule efficiency. Unfortunately, it is often unrealistic, because:

• It typically requires weeks or months of training before an agent can handle all possible interactions proficiently.
• Some interactions require a special personality, for example, sales and collections.
• Some interactions require certification that can take months or years to obtain.
• Staff turnover is a fact of life in every call center, so in many cases, the nirvana of a fully multi-skilled universal agent workforce can never be reached.

SBR helps centers to utilize staff in a way that may not be fully universal but is certainly more efficient than individual dedicated teams. New hires can be trained on the different call types one by one, progressing once they have demonstrated mastery and gained confidence by handling the interactions that are routed to them.

No free lunch

Multi-skilled agents and SBR present a number of challenges:

• Shift swaps need to be handled with more care. Unless agents swap with colleagues who have exactly the same skill combination, there will be an impact on customer service.
• Similar challenges apply to time off, overtime and sickness. In a single-skill environment, planners and team leaders can easily find a substitute for an agent who calls in sick. With SBR, the set of skills that the absent agent possesses really matters. If the missing agent has three skills, ideally you must find another agent with those same three skills to replace them. It’s easy to imagine a situation where contacts arrive but no suitably skilled agent is present to handle them.
• Last but not least, the complexity of the workforce management (WFM) process soars. Modeling multi-skill pooling efficiency in the forecasting and scheduling process is not a trivial task.

Staffing calculation

A fundamental challenge that SBR poses to WFM is that of determining how many agents are needed with each skill or combination of skills. In a single-skill or universal agent scenario, the Erlang C formula works well. When agents have a combination of skills, however, Erlang C is completely lacking, since it takes no account of agent skills. It assumes that all agents are identical. With SBR, the combination of agents and skills makes a big difference since there is a web of interdependencies. For example, hiring staff with skill X may not seem to address a shortage of resourcing in skill Y. But if there are agents with skills X and Y already working in the center, the new agents will free up agents with skill Y to take more calls.

Solutions

WFM software applications approach these challenges in a variety of ways. Some require the user to make simplifying assumptions and handle multi-skilled agents in the same way as single-skilled agents. Others use sophisticated simulation models that mimic the exact mix of calls in each interval, e.g. 15 minutes. They experiment with different shift combinations for the multi-skilled agents on a trial-and-error basis until a solution is found. Simulation is certainly powerful but does have some downsides:

• The simulations can take a long time to run, even in a small- to medium-sized center. That’s not only inconvenient, it makes experimentation less practical. More importantly, it limits the planning team’s ability to quickly respond when actual contact volumes differ from the forecast or when several agents call in sick.
• Simulation is extremely sensitive to the accuracy of forecasted volume and average handling time at interval level. That's because it models the arrival of different interactions in each interval. How accurate is your forecast? Many centers pride themselves on achieving ±2% measured over a day or week but they may have a variance that is nearer 10% at interval level. The lower the volume of a given interaction, the more likely it is that the variance will be high.
• Sickness and other shrinkage effects take place in every contact center. A simulation can't possibly predict which agents will be absent on a given day, and consequently, the validity of the schedules will be compromised.
• The simulation will inevitably use a simplified version of the truth. Schedules are typically built weeks in advance. When the day arrives, some agents may have left the business, new ones may have started, and others trained on new skills. Even a schedule for the upcoming week may need to be rebuilt based on shrinkage effects such as time off, team meetings, and training.
• What if you change the routing logic in the ACD or contact routing platform? Many centers make frequent changes to routing rules in response to changing business needs. These changes make the simulation results invalid and mean that you need to reprogram the simulation - a labor-intensive and time-consuming process.

Optimization

Another approach to creating schedules in an SBR environment is to use sophisticated optimization algorithms instead of simulation. Optimization offers a number of important benefits:

• Optimization builds schedules that take account of individual agent skills to an acceptable level of precision much more quickly than simulation. This speed enables timely, effective responses to changes in volume, call arrival patterns, or agent availability.
• Intraday adjustments can be calculated quickly and staff more precisely matched to demand when the actual workload and staff skill mix are known.
• The initial setup and ongoing maintenance effort are dramatically reduced. You don't need to update a complex configuration every time the routing logic is changed on the ACD.

While this approach may seem on the surface to be less precise than simulation, in reality, the optimization is completed quickly and typically results in schedule efficiency that is equivalent or even superior to that offered by simulation. Schedule efficiency is a key WFM metric. Also known as 'schedule fit', it is a measure of the extent to which the ‘supply' of agents matches the 'demand' for agents. The goal is perfect coverage, i.e. zero under-staffing and zero over-staffing. Planners have a clear advantage if they can constantly optimize schedule efficiency, quickly and easily, even in an SBR environment. The advantages of optimization are clear.

Conclusion

Generating efficient schedules in an SBR environment is a difficult task. The challenges don’t stop with schedule building. The planning team will be faced with operational challenges such as changing volumes, agent sickness, and business changes that affect average handling time (AHT). Any of these effects may require the schedules to be revisited. Remember, the main goal of WFM is to consistently put the right number of people with the right skills in their seats at the right time. SBR adds an extra level of complexity to this process, but savvy planners can overcome the challenges by choosing the right methods and tools.

▶️ Check out injixo’s multi-skill schedule optimization functionality