Employee Retention Statistics Your CFO Will Actually Trust

You’re in a budget meeting, asking for money to invest in your experienced employees and prevent them from leaving.

To make your point, you’ve brought the number everyone brings.

"It costs 200% of an employee’s salary to replace them," you say, quoting the figure you learned somewhere.

Your CFO looks up from her spreadsheet. "And who says that?"

Well, uh... your slide has a hyperlink to a blog, which links to another blog, which ultimately says nothing about which jobs the number applies to, what costs it includes, or whether the replacement employee’s salary is included.

Look, for a hard-to-fill specialist role your 200% figure might be accurate. But since your finance team can’t audit your number, they can’t defend it to someone higher up.

In fact, any number they can’t audit drops to zero in their book, and your request is instantly worth only a shot in the dark.

Sound familiar?

Most statistics about employee retention suffer from the same problems. They’re outdated, lack context about the methodology, or have been repeated so often that no one remembers the original study.

But this doesn’t mean you can’t make a business case for employee retention. It just means you’ve got work to do.

In this article, you’ll learn:

• How to find and present outside numbers your finance team will accept
• How to calculate a turnover cost based on your own hiring, payroll, and vacancy data that’s acceptable to your finance team
• How to account for offsets and edge cases that most HR models overlook

In this article

1. Why do finance teams get suspicious when HR teams talk about "turnover costs"?
2. Which outside retention statistics can you trust?
3. How do you spot the red flags in a retention statistic?
4. Do you know when your retention statistic has expired?
5. How can you build your own turnover cost model?
6. Which retention metrics should you track on a quarterly basis?
7. How can you turn your retention data into a business case your finance team will understand?

Why do finance teams get suspicious when HR teams talk about "turnover costs"?

From the HR side, turnover costs feel like a series of events that happen back-to-back and that you’re very familiar with.

On the Finance side, those turnover costs feel like a bunch of separate cost categories that have to be added up to come to an overall estimate.

Strictly speaking, Finance teams have a point when they say this. For instance, a manager’s time spent interviewing doesn’t add any new cash cost since they’re already getting paid. What Finance teams want to know is what other work did the manager stop doing to make space for these interviews.

Then you have costs that go down, such as when a job isn’t filled. The company doesn’t pay the employee’s salary for as long as the job isn’t filled. A lot of HR models don’t factor in these savings.

If your company can’t break down a turnover cost figure into its parts, or interrogate each part, then it starts to sound dubious.

Consider the example of the stat we mentioned earlier that says replacing an employee can cost companies anywhere from a sizable share of their yearly pay to well beyond it. If a CFO read this stat, they’d only take away that the type of role impacts the cost of turnover. They still wouldn’t know what the departure of your senior engineer last quarter cost the company.

Then there’s the question of where that number came from in the first place. Did your team glean it from a vendor’s blog? Did that vendor’s blog take it from another article?

Is the original study publicly available for anyone to check? Or was the study done in a completely different labour market in a different decade?

Vendors selling products to help companies keep employees have an incentive to make turnover costs sound alarming, so this may be part of why your finance team is skeptical. The same level of scrutiny should be applied to your own claims as well!

Finance departments don’t want HR teams to quit tracking retention - they just want to see clearly where the numbers come from and have the opportunity to question the assumptions behind them.

Which outside retention statistics hold up?

Before you plug a statistic from elsewhere into your business case, question it. Where did these numbers come from? How is turnover or cost calculated?

What time period does this cover?

With statistics, what you assume a number means and what it really means are often not the same thing. As we’ve mentioned before, there is no single, agreed upon definition for "turnover rate". Similarly, quit rates don’t account for layoffs and cost per hire doesn’t include the full cost of replacing someone.

Why does this matter? If you want to compare your turnover to the industry average, the most reliable source of turnover figures is national labour statistics. You want to be able to tell someone, "Is our turnover just the market?"

That said, pay attention to the nitty gritty details of how they define things. For example, in the US JOLTS data, retirements are classified as "other separations", not "quits". In addition, the JOLTS quit and separation rates come from data collected from establishments, which may not be the same data collection method as your internal figures.

Understand these differences before making comparisons.

You should also find data from the market or markets you actually operate in and see whether the data has been seasonally adjusted. Otherwise, you risk making inaccurate comparisons. For instance, a retailer that compares its January turnover to a seasonally adjusted industry turnover figure for the same month may paint itself in an undeservedly bad light.

When you find a stat to back up your recruiting business case, try to find information that explains how the numbers were derived. SHRM’s hiring benchmarks explain how their statistics were obtained. Their cost per hire accounts for both internal and external recruitment costs.

While SHRM figures can help you determine whether your recruiting costs are "high" or "low" compared to industry norms, it’s not a reliable figure for calculating the productivity cost of an empty job.

For that, you’re better off using a replacement cost estimate that depends on the type of role. Gallup publishes research like this. This is a better approach than using one multiplier for every employee.

According to Gallup, replacing a frontline worker costs a sizable share of their yearly pay, while replacing leaders and managers costs considerably more than their yearly pay. You can use this reasonable range to check whether your numbers make sense.

What if you want to create a business case to spend money on exit interviews? Is it enough to grab some general research on exit interviews?

Not quite. In fact, a large number of voluntary leavers provide a controllable reason for leaving. Work Institute’s retention research uses data collected from its exit interviews to produce the true reasons why people leave.

But remember that exit interview data is biased. For instance, an employee who still wants to use your company as a reference will likely provide "safe" reasons for leaving like pay or "a better opportunity" instead of offering his or her real, and possibly negative, experience with their manager.

So it’s important to remember that when you’re using these exit reasons, you can’t take them at face value and call them the "true" picture of controllable turnover. And you can’t simply plug these reasons into your business case and expect them to demonstrate how many people a new program will keep. If an employee says "pay", they clearly considered the hassle of looking for a new job and decided it was worth it.

How should I cite outside statistics?

If you’re pulling stats from an outside source, your finance department may quickly become doubtful. Here’s how to cite your sources so your business case garners the attention it deserves:

• Provide the original source of the report/survey and the publication date
• State the sample size or specify the particular roles or market the statistic is referring to or was drawn from
• Mention any particular details the statistic excludes

While this may seem like overkill for one slide, it’s less of a pain than having to defend a weak statistic. It also shows your finance department that you already question the numbers you use, which is why you carried out your own internal analysis.

If you regularly present business cases, it’s helpful to make up a simple rule for when outside data is good enough and when you need to produce your own data?

For instance, if your proposed spend is less than the cost of losing one or two employees in that role family, then strong, well-sourced data is good enough to get a pilot project off the ground.

If you’re asking for money every year as a regular line item in the budget, you’ll need to present your own figures.

How to spot red flags in retention statistics

Not all statistics that make it to your spreadsheet are worth using in your presentation. Before you use any retention stat, make sure you do a quick check for the following red flags:

• No primary source: You only have access to a press release or summary of a study instead of the actual study or data table? It’s hard to know what went into the statistic. A common example is a stat that includes the replacement employee’s full salary as a "cost" when you would have paid that anyway.
• Too clean, precise number with no range given: "3 times salary" lumps together jobs with significantly different vacancy times and recruitment ramp-up times. It’s worth asking questions about what cost categories were included and whether the vacancy salary saving was deducted.
• Uses an old labour market study: A study from before the widespread adoption of remote working - or from a time when the labour market had an unusually high demand for employees - can still be useful. But its time to fill and offer acceptance level assumptions aren’t reflective of a labour market that you’re currently trying to hire in.
• Lacks clarity on the type of turnover considered: Find out what the numerator and denominator are. Were all employees considered? Were voluntary, involuntary, or retirement turnovers considered? What about average headcount versus opening headcount? What’s the time period? Could the turnover rate be multiplied by four to get an annual figure when there are seasonal spikes?
• Analysing a different kind of workforce to yours: An interesting stat based on frontline roles would underestimate the cost of replacing a specialist who takes months to find a replacement. A stat based on specialists would overestimate the cost of high volume roles. Make sure there’s applicability between the study’s workforce sample and your company’s workforce composition before relying on a stat.

Even though you may want to use a weak statistic in retention, because everyone in the room recognizes it, you don’t want to run the risk of someone at the board level asking what researchers actually measured.

Apply the five above checks to your own internal statistics as well. If your HRIS system output includes contractors in your headcount or counts internal transfers as employees who left, that will cause issues when you use that data for analysis.

Be mindful of the limitations of borrowed statistics

Even if you plug borrowed statistics into a well-built study, you’re still using data that represents another company’s pay rates, hiring processes, and time to fill positions.

Your organization’s structure will also influence how quickly you can fill a position with a particular job title.

For instance, hiring internally saves you money on agency fees, but it may open up a second vacancy (unless you’re eliminating the internal candidate’s previous job). This is something benchmarking data doesn’t take into account.

If there’s a lot of internal movement within your organization, your external hiring costs may seem low, when in reality you’re accounting for long vacancies spread over two to three positions. Tracking where your internal employees move gives you a sense of these linkages and lets you view them as a cost of losing that employee.

As a general rule, you should only use benchmark data to check your numbers against. If your number falls very far out of a published range, investigate further. Oftentimes, this is a sign that a cost category was excluded or a particular salary was double-counted.

It may also be a true characteristic of your workforce that’s worth investigating.

As Douglas W. Hubbard points out in his book, How to Measure Anything, a range of estimates supports decisions better than a precise number that only seems reliable. Measuring your costs, even if it’s approximate, can reduce uncertainty enough to shift your decision making.

To make this impact, assign a low value and a high value to inputs you’re unsure about to see how your total cost estimates shift. Your finance team already uses this approach when they build forecasts, and your CFO will likely trust a range of figures more than a single number with no reasoning behind it.

The most trustworthy retention figure for your finance team is one your CFO can check against your own records.

Build your own turnover cost model

Start with the last 12 months of voluntary exits in one role family. Keep the first version small enough to check line by line.

Pull separation dates, requisition-open dates and replacement start dates. From recruiting, get agency fees, job ad spend, background checks, sign-on bonuses and relocation tied to those specific requisitions. From payroll, get accrued leave payouts.

Those are the hard costs. Add onboarding and training only where the spend is incremental per hire. Don’t load the full cost of a standing training program onto each new hire unless finance agrees that it scales with headcount.

Cost the vacancy in both directions

Next, cost the vacancy, both directions. Overtime, contractors and temp agency invoices go in. So does delayed revenue you can tie to operating data, such as unbilled consultant hours or a delayed project milestone.

Then subtract the salary and benefits you didn’t pay while the seat was empty. A model that skips that offset is the first thing a sharp FP&A partner will catch.

Check the net, not just the offset. If a contractor’s day rate exceeded the leaver’s fully loaded daily cost, the vacancy cost money even after the saving. If nobody covered the work and nothing measurable slipped, the vacancy produced a net saving, and you should show it rather than hide it.

Account for replacement pay and ramp-up

Then check the replacement’s pay against the leaver’s. If you had to pay above the departing salary to fill the role, the difference is a recurring cost the budget carries every year.

That premium also opens a pay compression question with the people who stayed in the same role. List it as a risk line under the total, not as a priced cost, unless you’ve already had to adjust their pay.

Manager time belongs in the discussion, carefully. Interview hours aren’t new cash. They’re work the manager didn’t do.

Agree with FP&A whether that sits as a labelled opportunity cost or outside the core figure.

Then the ramp-up gap. A replacement who starts on Monday isn’t at full output by Tuesday.

Ask the role’s manager to sketch a productivity curve, for example rough output levels at months one, three and six. Use production data where it exists.

Cost only the shortfall below full output. Write the assumption down and run a low and high case with finance.

Cost of turnover = separation costs + vacancy costs + replacement costs + ramp-up costs.

Keep cash costs, estimates and offsets in separate columns.

Check the edge cases

Several edge cases break a naive model. Decide how to treat each before you present:

• Roles that weren’t backfilled. If the seat was eliminated after the exit, there are no replacement or ramp-up costs, and the ongoing salary saving belongs to workforce planning, not to the turnover figure. Count their separation costs, but leave these exits out of the replacement and ramp-up averages.
• Boomerang rehires. A former employee who returns ramps up faster and costs less to recruit. Give them a shorter productivity curve rather than the standard one.
• Paid notice and handover overlap. If the leaver and replacement overlapped for knowledge transfer, you paid two salaries for one seat for that period. That overlap is a real cash cost.
• Counter-offers. A retention raise that keeps someone for six months before they leave anyway adds to the cost of that exit. Record it against the eventual departure.

Of course, one average won’t cover everyone. A warehouse associate and a senior engineer differ on every input. Build per-exit estimates by role family, then multiply by the exits you actually had in each.

Leave out what you can’t give finance a reasonable way to test, such as lost ideas or the awkward weeks for a team. A conservative model everyone accepts will win more budget than a complete one nobody believes. If you want to mention excluded costs, list them below the total as unpriced risk.

Bring FP&A in while you set the assumptions, not after the slide is done.

Which 5 retention metrics are worth checking each quarter?

Regrettable turnover

Regrettable turnover helps you distinguish between the employees you’d like to keep and those you can afford to lose. For instance, you may consider an employee essential to your business, doing a great job, and difficult to replace.

In a capable human resources team and manager team, the decision whether an employee is "regrettable" to turnover is a joint one. And it’s a decision that should be made ahead of time, well before an employee formally resigns, by looking at the employee’s role, performance, and other factors.

Further, keep in mind the criticality of a role when evaluating your regrettable turnover. As John Boudreau and Peter Ramstad argue in Beyond HR, talent investment should follow pivotal roles, where changes in the quality or availability of people have the greatest impact on business results. Prioritizing roles like this isn’t the same as focusing on the highest paid employees or the largest groups when setting goals.

Consider a mid-level employee whose job is pivotal to keeping a key revenue-producing process running smoothly.

First-year attrition

Unlike some of the other metrics on this list, you’ll have to gather the data yourself for your first-year attrition.

Create cohorts by listing every employee you hired during a given time period, then track what happens to them during their first year. Within this cohort, you’ll want to separate employees who leave within 90 days from those who leave between 90 days and one year.

This is helpful because it breaks down opportunities for improvement. People who leave within 90 days are often a symptom of poor hiring decisions or jobs that don’t match previews. Employees who leave after 90 days but before reaching a full year of service may not be receiving a proper onboarding experience or may be unhappy with their manager.

When reporting on this metric, it’s important to show both the raw numbers and the percentages, since small cohorts can skew the results. For example, if you have 25% attrition rate in a cohort of 8 (meaning 2 people left within the set time period) then the 25% isn’t very meaningful unless you know what happened to those two people.

Turnover by manager

Tracking your turnover by manager helps you understand what may be causing clusters of turnover within your overall turnover rate.

For example, you may notice turnover within a specific department, but it could be due to a different manager or work setting.

This is a metric that’s easy for companies to misunderstand or get the wrong message from, so be mindful. Use rolling 12-month periods, and take into account when a manager took over a new team or when a reorganization happened. From there, you can compare managers with similar types of work to one another.

For instance, it makes sense to take a closer look if a manager with eight people on their team loses three.

Cost of turnover by department

Once you’re comfortable putting a value on employee turnover, it’s worth looking at the cost of turnover by department.

You may be surprised to learn that the departments with the most employee separations aren’t necessarily the most expensive for the organization. For instance, if you have specialist roles that take several months to fill, the department with the most employee separations may not be costing the organization the most.

Present both rankings to department heads so they notice the difference.

Retention of high performers

If your organization has a consistent, calibrated view of what makes someone a high performer, it’s helpful to have a retention of high performers metric that compares overall attrition to high performer attrition.

This shows you whether you’re losing more than your share of high performers or whether they’re staying at about the same rate as everyone else.

Keep in mind that there is a lag effect with performance ratings. These ratings come from past performance cycles, so a high performer who’s already made up their mind to leave will still be rated as a high performer on the day they resign.

And don’t assume that every high performer wants the same things out of their job.

While surveys help drill into these findings, they don’t prove cause and effect.

How Sparkbay can help you identify retention risks before it hits your bottom line

With Sparkbay, you can identify retention risks before it hits your bottom line.

Instead of waiting for turnover data - which only appears after an employee has left - Sparkbay helps you get in touch with your employees before it’s too late. Using short surveys, sent out on a regular cadence, you’ll have a pulse on how your employees feel.

These data points allow you to keep your finger on the pulse of different areas of your business. For example, if you notice an uptick in turnover in one area of your business, you can look at the engagement data for that location to see if there was a sharp decrease in engagement that preceded this turnover, or if the turnover happened out of the blue.

Since you can customize the wording of surveys and dashboards, you can gather the data you want to focus on the areas your turnover data points to.

Sparkbay allows companies to compare specific groups (e.g. departments, managers) side-by-side. This is helpful for large organizations where the overall, company-wide average engagement score may be masking problems within a specific team.

Sparkbay automatically sets up report access based on your organizational hierarchy, so every manager only has access to data for their teams. In addition, to protect employee anonymity, engagement results are only displayed after a minimum number of responses are received (5 by default, but customizable).

If you’re interested in learning how Sparkbay can help you build a more engaged workforce, you can click here for a demo.

Turn your retention data into a finance-friendly business case

Once you’re in the meeting, lead with the number that your model found.

"Okay, so we’re losing X number of people from Y role families, and I’ve identified that our direct costs based on our spend records and some estimation came to $Z amount."

At this point, it’s worth specifying that you want to reduce regrettable turnover.

Why? Because some employee turnover is healthy for your organization. Making a business case that says you want to reduce ALL employee turnover may make other stakeholders question whether you’re also trying to keep employees you’d like to part ways with.

Once you’ve asked for money to address a specific problem area, direct your proposed intervention.

Are you spending a lot of money because new hires are leaving within the first year? You may want to test a more streamlined process for handing off candidates from the recruiter to the hiring manager, and then to the onboarding process.

Does one group of employees keep re-opening their requisitions? Understand the issue within this group before proposing a company-wide onboarding program that may spread your resources too thin.

As part of your business case, present your cost-benefit analysis in a way that’s easy for others to challenge and discuss. You can do this by presenting the core math of the analysis on the slide. For example, present the break-even cost equation:

Break-even exits = cost of the intervention ÷ per exit cost in the target role family

Make the number of break-even exits clear in your business case, and include information about how long the program will take to prevent those exits. And if possible, include a conservative scenario that can be viewed in comparison to the main business case.

This way, if someone is sceptical of your assumptions, they can fall back on the conservative business case. A break-even number of two or three prevented exits is more compelling for a CFO than a grand projected savings.

Use outside research to validate why this is a problem worth addressing, but use your own data and records to make a direct ask for money.

Finally, set a review date and agree on evaluation metrics with the finance department.

Ideally, incorporate a leading indicator (e.g., employee engagement in the pilot group) since a 12-month turnover rate may not change within this 6-month review window.

If at all possible, identify a comparison group that didn’t receive your intervention. This is especially important if you’re planning to roll out an intervention in your worst-performing team.

You selected this team based on its metrics, which are not ideal. So there’s a chance your team will naturally revert back to the mean over time. Without a comparison group, it’s easy to credit your program for this natural reversion to the mean.

If you’re interested in learning how Sparkbay can help you build a more engaged workforce, you can click here for a demo.