Contact Center Forecasting Fundamentals Part 2: How to Master Workforce Forecasting

In part 1 of our Contact Center Forecasting Fundamentals series, we explored various ways to forecast demand (or workload), in the long- and short-term. Now it's time to forecast the available staff in your contact center.

This is important, because only when you know how many staff are needed and how many staff you have available, can you take steps to align supply and demand. (I will explore techniques that you can use to close the staffing gap in part 3 of this series).

Forecasting available staff

Step 1: Find the total number of employees

The first step is to determine how many individual employees you have available.

Pro tip: Use your company’s HR system to find the number of employees currently available in your contact center. That should always be the starting point.

Be careful only to count the employees you intend to schedule. Typically this will be the agents, but some workforce management teams also schedule coaches, team leaders, etc. It’s important to ensure that the agent count in the HR system matches that in the WFM system. If not, take steps to reconcile the two so that there is ‘one version of the truth’. You should reconcile headcount at least monthly, if not weekly. This is the foundation for your staffing forecast.

Step 2: Calculate full-time equivalents (FTEs)

The staffing requirements we determined in part 1 of this series (based on workload and service level) are counted in FTE. Therefore, the calculation of available staff should also be based on FTE. That enables us to compare supply and demand on a like-for-like basis.

Full time vs. part-time

In some contact centers, everyone works full time. In this case, it's not necessary to adjust for part-time employees. However, if you have employees who work less than full-time hours, you must take this into account. The definition of full-time hours varies by region. For example, in the United States, a full-time employee typically works 40 hours per week.

There are two main ways to factor in part-time workers:

1. Count all part-time workers as 0.5 FTE

This is the simplest method and is appropriate if you only have a small percentage of part-timers or if you don’t have the bandwidth to keep track of the actual working hours of your part-timers. It’s less accurate but gives you a directionally correct number so you don’t overstate your workforce.

2. Calculate part-time workers based on their contracted hours  

This is the more accurate method and is straightforward to calculate. For each part-time employee, take the number of hours they are contracted to work and divide by the number of hours that make up an FTE in your contact center. For example, if someone works 24 hours and a full-timer works 40 hours, divide 24 by 40, i.e. 60%. This person would count in your staffing model as 0.6 of an FTE.

Step 3: Adjust for attrition

Staff attrition is a fact of life in practically every contact center. When establishing the number of available FTEs, you need to take into account that a certain percentage of them will leave within the period for which you are planning, and reduce the count of available staff accordingly.

Many capacity plans only track 'attrition' in general. However, this mixes two very different reasons why employees leave the company. Voluntary attrition is initiated by agents and involuntary attrition is initiated by the contact center. If you record these separately, you can better track the different reasons for attrition and thus more accurately predict the staff that will be available in the future.

Voluntary attrition

Voluntary attrition occurs when an agent leaves the company by choice. When you plan voluntary attrition, you should look at historical data to identify trends. Much like when forecasting volume, you should look at how the rate has changed year-over-year and also analyze seasonal influences.

You may notice that attrition goes up after people receive their end-of-year or quarterly bonuses. You may also notice a change in voluntary attrition based on the tenure of the population. New hires (those who worked in the center for less than a year) generally have a higher attrition rate than employees who have been with the company longer. So you should determine the voluntary attrition rate for new hires separately from tenured employees.  

A good way to build the model is to look back at the last several years. In my experience, 36 months is a good period to evaluate, but it varies by industry and depends on how stable the attrition rates of your business are.

Pro tip: For the calculation, determine the average attrition rate for the first employee group (i.e. new hires) for January, by looking at the January data for the last three years. Perform the same calculation for each month of the year. Then, repeat everything for the second employee group (i.e. tenured employees).

In order to calculate the attrition rate for all employees, use a weighted average based on the percentage size of the respective group. Let's look at an example.

Example: Attrition rates for January 

1. Calculate the average attrition rates in January for the two employee groups.

2. Determine the weighted average for all employees in January.

Repeat this process for each month. It’s a simple set of calculations that you can easily do in Excel. Start with the formulas for January and then copy/paste for the remaining months.

Involuntary attrition

Involuntary attrition takes place when employees are terminated by the company, usually for violating policy or because they don't meet the requirements of the performance management processes.

The violation of policies and poor performance are two different reasons for the dismissal of agents. To calculate future attrition rates correctly, both reasons must be considered separately. For example, policy guidelines may change over time or quality standards may rise. You should determine the impact of both changes separately, too.

On top of that, the attrition rate is also affected by whether or not your company actually enforces the new policies.

Example
Imagine your company has an absenteeism policy that states that an agent will be terminated after three unapproved absences in a single month. Then, the policy is changed so that two unplanned absences result in termination.

Your historical data is based on the old policy, but there will be a new policy that is more aggressive in the future. You might expect that the number of terminations will rise and consequently, so does the involuntary attrition. But that won’t be the case if the policy is not actually enforced.

This is where you really need to understand why the policy is in place (or changed in this case), and how operations management plans to enforce it. In this example, if the objective is to only send a warning message, but not to actually terminate more people, your attrition rate won't change. Even if they do intend to enforce it, you should check the number of employees who had unplanned absences twice a month vs. three times a month. If the number is the same, then essentially the policy won’t result in more terminations and the attrition rate will stay the same.

As you can see, the work to determine involuntary attrition can be complicated and requires a lot of engagement with your partners in operations. It also requires interpretation. But, the payoff is huge in terms of accurately estimating the number of available staff.

Top 3 mistakes to avoid when forecasting available staff in your contact center

1. Not reconciling headcount

Headcount reconciliation between HR and planning should be done regularly. Without an accurate understanding of exactly how many people you have, you don’t have a solid baseline to work with.

For scheduling, you should reconcile weekly. In this series, we are focusing mainly on long-term forecasting, so monthly headcount reconciliation will give you what you need to identify gaps for your hiring and training plans.

2. Not properly accounting for attrition in small agent groups

Attrition is challenging to forecast, especially in small organizations. If you have 50 people in a staff group and project 1% attrition per month, that’s a half person.

Do you reduce the count of available staff by 1 person or 0? Neither is technically correct and making the wrong choice will impact whether you’re overstaffed or understaffed. 

Pro tip: if your contact center focuses more on customer satisfaction, err on the side of caution and round down, i.e. assume you will have 49 agents. On the other hand, if your main focus is on cost reduction, assume you will have 50 available agents.

3. Not properly separating agent groups

In contact centers, there are types of calls or contacts (e.g. sales, complaints, service) and there are groups of agents (teams). Usually, each agent group handles a specific type of contact. Agile contact centers use tactics like overflow, i.e. forwarding calls to other teams in case of an emergency. When this happens, agents who normally receive, e.g. service calls, now also receive calls from another call group, e.g. complaints. It can also happen that entire agent groups are temporarily made responsible for a different call type.

If complicated overflow rules are used in your contact center or entire agent groups temporarily take over other call types, it is easy to lose sight of which agents belong to which group.

Remember that workload and staffing requirements are typically forecasted for specific call types. Therefore, it is important to also group agents by the call types they normally receive, not the call types they are currently handling for short-term operational reasons.

Always be well-informed 

Are you informed when an employee moves from one team to another? Does management tell you when a change is upcoming? Are all changes entered promptly into your WFM tool? If you want to be at the top of your contact center forecasting game, the answer to all these questions should be "Yes".

It’s very important for planning that you are informed when a team is suddenly responsible for a new call type or an employee changes from full-time to part-time. You should also be informed of all new hires, transfers, and terminations. Much of this information can be obtained directly from the HR system.

It’s crucial to define a clear process for when you want to be informed, by whom, about what.

Conclusion

Forecasting the available staff in your contact center is key to a robust staff plan and often doesn’t get the focus it deserves. It’s worth spending time to do this right.

You should have regular meetings on this topic, and if it doesn’t take up a significant amount of your (or your team’s) time, then you’re probably not spending enough time on it. 

It’s also important to note that the forecasting error for staffing availability can be large, because you rely so heavily on insights and observations in addition to historical data.

That's why you should measure your historical accuracy and keep the accuracy level in mind as you’re making decisions. If your forecast accuracy for the available staff is +/- 1%, you can build important business decisions on it.

If the accuracy is +/- 20%, then it’s just guidance, and you have to assume a high probability that actual staff numbers will deviate significantly from the plan in the future. If that is the case, you’ll need more staffing flexibility to react to over- and understaffing (overtime, voluntary time off, schedule changes, etc.).

Up next

The third and last post in this series offers best practices to close the gap between staffing requirements and available staff. I’ll share some proven techniques for effectively aligning staffing with demand, both in the short and the long term.

Part 3: Contact Center Forecasting Fundamentals: Closing the Staffing Gap