What’s Your True Cost of Attrition?
A Guide to Identifying and Calculating the Hard
and Hidden Costs of Losing Employees
By Ron Davis, Tenacity
An Executive’s View
As the CEO of a company that helps contact centers reduce employee turnover, I have a lot of conversations with executives about the cost of attrition. Their estimates are as random and dangerous as a game of drunken darts. I find this especially amazing, because year after year these leaders fill out surveys saying that agent turnover is their #1 problem. But after decades of hand wringing, they have no idea what it costs them. For an industry dedicated to painstaking measurement of employee performance, this is surprising. And for an industry suffering with razor thin margins whose biggest preventable costs come from agent turnover, it’s inexcusable.
It’s true that many of these senior managers think they know what it costs when employees leave. Perhaps they read an interesting article about agent retention online, or a consultant gave them a rule of thumb. Maybe they were a bit more ambitious and got someone from the finance team to try and model the cost of agent attrition in a particular call center four or five years ago. Or someone sat down and did some back of the envelope calculations to figure out the hiring and training costs, assuming that’s most of the cost of agent turnover. If only they knew
In most cases, when I dig a little deeper, I learn that these call center bosses have very little idea of the actual, hard, measurable, bottom-line costs of losing their employees. As an example, we recently spoke with a senior executive about employee retention at his North American contact centers. His thousands of agents go through six weeks of training and two weeks of heavily supervised calls afterward, and then have a nearly eight-month learning curve before becoming fully productive. He said the average cost per lost agent is around $3000. After asking a few more questions, it became obvious that the real, hard, measurable, tangible cost to his bottom line was a bit more than four times as much.
Why is Measurement So Poor?
The reason for this variance is twofold. The first is that the industry has no widely held best practices for measuring the cost of employee turnover, and none of the thought leaders seem to have dedicated enough mindshare to change the way the industry thinks. And fixing this requires more than careful intellectual work – it requires leadership. Unless someone drives the industry forward to embrace standard forms of measurement, the drunk dart “measurements” will continue.
The other reason is incentives. If 5% of your employees quit each month, and there is no standardized definition of the cost of attrition, would you rather report to your boss that this costs $3000 per person, or $12,000? Mark Twain said there are “lies, damned lies, and statistics.” Clearly, he had never seen a financial model designed by the person whose performance would be judged by its outputs. Want to fix attrition? Start by getting honest with yourself about its costs.
It is easy to criticize. Instead, I’m going to outline the items that should be in any model that estimates your call center turnover costs. Building from this, executives should be able to figure out how much agent retention is really worth. That way, when they make cost-benefit analyses about employee retention efforts, they aren’t playing make believe.
Defining and Segmenting Attrition
The first task is defining agent attrition. What qualifies as an “attrit?” Some companies do a good job of measuring voluntary (agents quit) versus involuntary (you fired them) agent attrition. But there are often misclassifications. For example, at one leading BPO, when an employee just stops showing up to work, he gets classified as an involuntary attrit, because he technically got fired for not coming in. But clearly he quit; the decision was his. Figure out the primary reason they leave: you decide they’ll leave or they decide they’ll leave.
This matters a great deal. Perhaps you lose 12% of agents in the first month after they are released into training. If only 2% are voluntary, and 10% involuntary, your diagnosis and treatment will be very different than if these numbers were reversed.
I recommend bucketing agent attrition into at least two groups. First, track turnover for agents once they start making unsupervised calls. And then lump everyone who hasn’t yet reached that level of responsibility (trainees) into a separate group. The cost of attrition is much different for the two groups, and sometimes the solutions to the problem are very different too.
You can also get more granular. Track for each stage: the hiring stage, the training stage, the supervised calls stage, the first week, month, and quarter of making calls, as well as the time when agents reach the top of the productivity ramp and become “fully baked.” If you don’t know this already, you will soon see that the longer agents stay, the more likely they are to continue to stick around.
Calculating Rate and Costs
This is the easy part. For any of the buckets, take the number of agents who quit and divide it by the total number of agents in the bucket. If three out of twenty trainees quit, that’s a 15% rate. And divide into voluntary and involuntary whenever you are able.
Once you know the rate, the next step is to attribute a cost of attrition. Often when I speak to call center executives, hiring (and training) costs are the only items they include in their tally of the costs of employee turnover. This is a mistake. Still, it’s worth recounting all the pieces that should be included.
Say, after a hard day’s night shift, George quits, after one year with the company. There are certain predictable rhythms that follow: HR has to process him, revoke his access to the building, close out payroll, etc. Since HR does this so frequently in contact centers, they are pretty efficient at it. Still, you should count the time and paperwork costs. These are your separation costs for losing George.
Next up is sourcing his replacement(s). Perhaps you use a recruiter who charges a fixed fee. Or perhaps your recruiter charges a percentage of salary for the first six months. Or perhaps you source internally, through advertising, combing through applications, interviewing, background checks, and the like. Whatever the costs, figure out how many people you hire and divide by the total costs to find the average.
To keep the math easy, let’s say you hire a freakishly small class of three people and you typically only have one in three people make it through the full training cycle. John, Paul, and Ringo come together to replace George and they cost $500, $1000, and $1500 because they come from different sources. Divide by three and your average cost sourcing a new call center agent is $1000.
Here is where HR processing returns. John, Paul, and Ringo need to be processed. They must be put in your system, added to payroll, given a corporate ID, and security access. This includes any reporting to the state or employment bureaus and anything else that needs to be done to get this person into his first day of training. As with separation, figure out the total cost and divide by the number of hires. This is your hiring cost per person.
The cost to train new agents is a major source of legitimate variance in the cost of employee turnover. Some centers require only a few days of training. Some require months. We find the average to be around three or four weeks of classroom training.
Here you want to measure the marginal costs, which are the additional costs each time you train someone new. Since you are always going to have some employee turnover, you don’t necessarily need to amortize the fixed costs from your basic training infrastructure.
These include the cost of holding a training class. First, figure in the total bill to pay the instructors. This includes salaried and hourly employees who are taken away from other productive work. Next is the price of software and materials. If any of your training materials or technology is charged on a per user or per class basis, you should add that to the total tally as well.
The math on this one is pretty easy. You add the above to get the total cost per class and you divide by the number of successful graduates. This way, if Paul decides to run for his life and drops out, you have baked in the costs associated with his training into your final product (John and Ringo). After all, training Paul is part of the cost of replacing George. If the cost for the class, including Paul’s costs, was $3000, then the cost per successful trainee is thus $1500.
This is a big line item. Here you have to include other salary related costs such as benefits and taxes. If John, Paul, and Ringo makes $10 per hour, their true salary cost will be more like $12 per hour. Multiply this by the number of hours it takes to finish the training. This is the salary cost for your successful graduates. For each graduate in a 4-week class, this is 160 (hours) x $12 (wages/overhead) = $1920.
Defining and Calculating Leakage
Another mistake executives make is to ignore leakage. They only count the cost to get an agent through training. They tend to think like this: George quit? Now we have an empty seat. John takes his place. The cost to get John there was X. But they forget that it takes more than one trainee to fill that seat (can’t forget Paul and Ringo!). A lot of agents quit during the hiring and training process, and this is expensive. It has to be figured in too. If you lose 25% of trainees during training (leaving you with 75% of the original hires), that means you have to hire 1.33 trainees to fill George’s empty seat, and you can’t ignore the costs of that other one-third of a person.
- Take the TOTAL salary and overhead costs for the entire class, including everyone that walks out the door, rather than the individual costs, and then divide by the number of SUCCESSFUL graduates.
- Figure out the salary and overhead during training costs per successful graduate (say, $12 x 160 hours for a four-week class = $1920). Figure out the average rate at which people quit training (say, 30%) and average percentage of the training they finish (say 50%). Multiply $1920 x .3 and .5. This adds $345.60 to the cost per successful graduate.
If you are still unsure, think of it this way. Remember that George quit at the beginning of our story, and we are filling in his spot. We are measuring the cost of George’s attrition. Paul quits during training. John and Ringo haven’t yet (we will soon see that John is the one that makes it all the way through to the floor). The cost of replacing George (in this case) includes John, Paul, and Ringo’s time in class. The lesson here is that in order to calculate the cost of agent attrition, you don’t calculate the cost of the replacement agent, because that leaves out all the ones that don’t make it to the floor. You calculate the cost to fill the seat, which includes the Pauls and Ringos of the world.
There is a “halfway house” stage immediately following training. It’s when agents start answering calls, albeit at a grossly diminished rate and with a lot of supervision and feedback. This usually lasts for a week or two. John and Ringo may have been stars in your class, but this is where they are really put to the test: talking to customers. In the halfway house, there are three major sources that contribute to the cost of agent attrition: support, wasted salary and software, and, like training, leakage.
Support is the cocoon that you wrap around your budding agents. Usually this comes from training personnel. Imagine that John and Ringo have a little help from their coach, Janice. If Janice coaches just these two for the week, but Ringo quits early, all of Janice’s salary ultimately went toward only John graduating. Any teaching or support associated with the halfway house should be included here. It should all be added up and divided across the universe of everyone that graduates. In our example, the only one that makes it through this (tiny!) class is John.
Lost Salary Costs
The next cost is wasted salary. In the halfway house, agents receive wages but are extremely unproductive, answering very few calls. Using your workforce optimization software, you should be able to determine how they compare, on average, to a fully baked agent that has finished the productivity learning curve (usually about six months after training).
Let’s assume here that halfway house agents are completing 1/3 as many calls. That means 2/3 of John and Ringo’s salary and technology costs and overhead is wasted. Let’s say the software licenses cost $3 an hour, bringing their total per hour cost to $15. So for John, at $10 per hour in waste, that’s $800.
But then there’s that sneaky leakage problem again. Ringo quits after a week, so he wastes $400 in wages. Since they are the whole class, you add these together ($1200) and divide across whoever made it out — in this case, just one person, John. Now we have filled George’s seat with John, and the cost of the halfway house salary waste, including leakage, was $1200.
Ramping Up and Bulking Costs
There are two more major sources of cost from agent attrition, and they are two of the most undercounted line items. First is the productivity ramp. Most call center managers know how long this is. During the first few months on the floor, agents aren’t on average very productive, but they improve pretty significantly. At some point (often about six months), this levels off. At that leveling off period, we call them “fully baked.”
The time between the end of the halfway house and being fully baked is the productivity ramp. Recall that our friend George quit his job after a rough day. John, Paul, and Ringo got hired as replacements, and Paul quit during training and then Ringo quit during the halfway house. Now John is sitting in Ringo’s chair, answering calls on his own.
The productivity ramp can be quite expensive. To keep the math simple, let’s assume that the first month an agent is 40% productive, the second month he is 50% productive, and so on until after the end of the sixth month, he is 100% productive. That means that in the first month, 60% of his salary, overhead, and marginal technology costs (i.e., for programs that charge on a per seat basis) are wasted. If he costs $15 per hour, that means $9 per hour is wasted the first month ($1440), $7.50 per hour the second month, $6 per hour the third, and so on. The total waste for John is $5040 over 24 weeks. While this ramp is steeper than most actual ramps, this is usually a very large line item, adding about 35-45% to all the previous costs of agent attrition we have discussed.
The last major item is also the hardest to wrap your head around. I call this “bulking up.” More attrition means a less tenured workforce. Because a workforce is, during its productivity ramp, less productive, that means it can’t answer as many calls. So not only is part of the salary of the new agents wasted during the ramp, but you also have to hire more agents just to produce the same output.
Doing the math on this one is more difficult, and you may need some assistance and data from the financial department. First, you will need to make sure you have a good grip on your productivity ramp. What is the average percentage of productivity for each of the months during the productivity ramp? And then see how many agents are in each of the first six months of production.
To keep our example relatively simple, let’s figure your average agent after the ramp is taking 100 calls a day and that you have 5% monthly attrition and that everyone makes it through the ramp. If 5% are in month one and answering 40 calls a day, 5% are in month two and answering 50 calls a day, 5% are in three answering 60 calls a day, and so on, you end up with 30% of your agents answering an average of 65 calls a day, and 70% of your agents answering 100 calls a day. This means your average agent is answering 89.5 calls.
Assume your call volume runs 100,000 calls per day. If you had zero attrition, you would need 1000 agents. But because you have 5% attrition, you need 1117 agents. Obviously you cannot and do not want to have zero attrition. But run the same scenario with attrition one point lower. Then you have an average of 91.6 calls per agent, or 91,600 calls per day. In this scenario, you need 1092 agents.
To then calculate this last line item in the costs of attrition, look at the difference between the 4% and 5% monthly rates. One percentage point of difference per month means you have to have 25 more agents on staff per year. If each agent, on average, costs $30,000 per year in marginal costs, that means 1 point of monthly attrition costs an additional $750,000. Since a 1% attrition rate per month in a center with roughly 1100 agents equals 132 agents per year, that means that “bulking” adds $5600 per agent that quits.
Now these numbers were simplified and it inflates the total. Your agents are probably better than 40% productive in their first month and their ramp is faster at first. A more typical, actual tally in a four-week training scenario, with a week or two of halfway house time, and a six-month ramp, is a bulking cost of about $2600. In any case, this is a huge line item that call centers tend to ignore. Even though it takes more work to measure, it is a real hard cost.
That’s the primary list. There are others that are worth considering in certain circumstances, although they are harder to measure. These include BPOs aiming for rewards or avoiding penalties from clients or trying to win clients. All benefit from lower attrition. But these risk-based calculations are more complex and contract dependent. Captured centers that want to avoid outsourcing face similar risks, but these are harder to quantify, unless there are known internal triggers for outsourcing. And CSAT and NPS are significantly affected by attrition, but it is very difficult to put an exact dollar value on these.
I hope this journey has been educational. If your centers have a three or four-week training process, you pay people 10 bucks an hour, and you have even normal leakage in the training and halfway house and a six-month productivity ramp, you now know that your “it costs $5000” number is just a pants-on-fire statement (it’s roughly $10,000).
The only executives I meet that have really genuinely been able to tackle attrition are the ones that know how to define and quantify it properly.
To learn more about the costs of attrition along with strategies and tools for reducing it, visit gotenacity.com.
Ron Davis is CEO of Tenacity, a company specializing in retention solutions for today’s contact centers. Tenacity uniquely combines social sciences, stress management, and resiliency best practices to deliver practical solutions to human performance problems. Contact Ron at email@example.com.