Most small businesses that outsource HR discover the same thing about six months in: they have no idea whether it’s working. The invoices arrive, payroll runs, nobody has complained loudly, and that becomes the whole evaluation. It holds up right until a tax notice shows up or a good employee leaves and the exit interview points at something the provider was supposed to be handling.
The fix is not a longer contract or a more detailed statement of work. It’s a short list of numbers reviewed on a fixed schedule. Twelve metrics, grouped into four categories, reviewed quarterly, is enough to tell you whether an HR outsourcing arrangement is delivering. This article gives you the formulas, the benchmarks worth measuring against, and a scorecard structure you can run in a one-hour quarterly review.
One note before the list. Many articles on this topic recycle software and subscription metrics like MRR, ARR, and churn rate, which measure a vendor’s revenue rather than your outcomes. Those numbers describe your provider’s business, not your HR function. Everything below measures your side of the arrangement.
The 12 HR Outsourcing Metrics Worth Tracking
Group them into four categories. Three metrics each is enough to see a pattern and few enough that a small business owner will actually review them.
Category 1: Service delivery
| Metric | Formula | What to watch |
|---|---|---|
| SLA compliance rate | (Requests resolved within SLA ÷ total requests) × 100 | Below 95% consistently means the SLA is decorative |
| Median resolution time | Median days from ticket open to close | Use median, not average. One 40-day case distorts a mean and hides the typical experience |
| Rework rate | (Transactions requiring correction ÷ total transactions) × 100 | The most under-tracked metric here. Rework is where outsourcing savings quietly disappear |
Category 2: Payroll and accuracy
| Metric | Formula | What to watch |
|---|---|---|
| Payroll accuracy rate | (Error-free payslips ÷ total payslips) × 100 | Target 99%+. Every error costs you employee trust disproportionately |
| On-time payroll rate | (Pay runs completed on schedule ÷ total pay runs) × 100 | Should be 100%. Anything less is a contract conversation, not a metric trend |
| Agency notices received | Count of tax or regulatory notices per year | The cleanest external signal of filing quality. Should trend toward zero |
Category 3: Talent outcomes
| Metric | Formula | What to watch |
|---|---|---|
| Time to fill | Median days from requisition approval to offer accepted | Track median by role type. A blended average across junior and senior roles tells you nothing |
| Cost per hire | (Internal costs + external costs) ÷ number of hires | Use the SHRM/ANSI standard formula so your number is comparable to published benchmarks |
| Voluntary turnover rate | (Voluntary separations ÷ average headcount) × 100 | Separate voluntary from involuntary. Combining them hides the signal you actually need |
Category 4: Employee experience and risk
| Metric | Formula | What to watch |
|---|---|---|
| eNPS | % promoters minus % detractors | This is eNPS, not customer NPS. Different question, different benchmark range. Don’t compare against customer NPS figures |
| 90-day new hire retention | (New hires still employed at day 90 ÷ total new hires) × 100 | The sharpest early indicator of onboarding quality |
| Open compliance findings | Count of unresolved findings from the last audit | Age matters more than count. A finding open past 90 days is a different problem than a new one |
Two rules that matter more than the metric selection. Establish a baseline before the provider starts, because without a pre-outsourcing number you cannot attribute any improvement. And keep the definitions fixed, since the most common failure is a provider quietly redefining what counts as a resolved ticket in month seven.
For benchmarks, pull from your own segment rather than national averages. SHRM’s recruiting benchmarking data is the standard reference, and its 2026 brief found median cost-per-hire for executive positions rose to $15,000 from $10,600 the previous year, while median time-to-fill for nonexecutive roles actually decreased. Two metrics moving in opposite directions in the same year is exactly why a single blended national figure is close to useless for a small business. Segment by role type or don’t benchmark at all.
Measuring Cost Efficiency Correctly
The comparison that matters is your fully loaded per-employee cost of HR before and after outsourcing. Not the invoice, and not the provider’s pricing model.
Fully loaded in-house HR cost per employee per year includes the HR salary and employment overhead, HRIS and payroll software licensing, compliance and legal consultation, recruiting tools and job board spend, training, and the founder or manager hours absorbed doing HR work informally. That last one is almost always excluded and is almost always the largest hidden line for businesses under 30 employees.
Fully loaded outsourced cost per employee per year includes the provider fee, plus the internal hours still spent managing the relationship, plus rework cost, plus anything carved out of the contract that you still pay for separately. Benefits brokerage and workers’ compensation are the two most commonly assumed-included and most commonly not.
Divide each by headcount and compare per employee per year. Then track the trend rather than the snapshot, because the interesting question is not whether outsourcing was cheaper in month one. It’s whether the gap holds as you grow, and for many arrangements the per-employee advantage narrows above a certain headcount.
One structural warning worth building into the contract review: per-employee-per-month pricing means your costs scale linearly with hiring, which is fine while you’re small and less fine at 60 employees. Ask what the pricing looks like at double your current headcount before you sign, not after. The same tension shows up when distributed teams weigh employment models, which is covered in more detail in this comparison of global payroll services versus contractor management.
What the PEO Research Actually Says
If you’ve researched HR outsourcing, you’ve encountered a specific set of statistics: businesses using a PEO grow at more than twice the rate of comparable companies, have 12% lower employee turnover, are 50% less likely to go out of business, and see an average return on investment around 27%.
Those figures come from research commissioned by NAPEO, the industry’s own trade association. That doesn’t make them wrong, and the underlying work covers a large sample. But it does mean they’re advocacy research, and there are two things worth understanding before you use them in a decision.
First, the selection problem. Businesses that choose to engage a PEO are not a random sample. They’re businesses with the cash flow, growth trajectory, and management maturity to sign a PEO contract in the first place. Some of the growth and survival advantage in that data belongs to the kind of company that hires a PEO, not to the PEO. The studies control for company size and industry, but not for the underlying disposition.
Second, and more usefully, look at the actual turnover numbers rather than the relative claim. The NAPEO research reports annual turnover of roughly 50% among PEO clients against roughly 58% among non-clients. The headline is “12% lower turnover.” The number that should change your planning is that half your workforce still leaves every year. If you engage a PEO expecting retention to be solved, that expectation is not supported by the industry’s own most favorable data.
The right use of these figures is directional. HR outsourcing plausibly improves administrative reliability and access to better benefits, and both plausibly help retention at the margin. It does not fix a retention problem caused by compensation, management, or the work itself, and no provider metric will tell you which of those you have.
Employee Satisfaction and Engagement Metrics
Employee satisfaction is a significant indicator of HR outsourcing success. The Net Promoter Score (NPS) can be used to assess how likely employees are to recommend the company’s workplace. A high NPS indicates strong employee satisfaction and can reflect positively on the effectiveness of the outsourced HR services.
Customer Lifetime Value (CLV) isn’t just for customers; it can be adapted to measure the long-term value of employees, factoring in recruitment, training, and retention costs. By analyzing CLV, businesses can determine if their HR strategies are fostering a productive and loyal workforce.
Furthermore, a well-designed onboarding workflow is critical for new employee integration and long-term engagement. Tracking metrics related to onboarding can help businesses ensure that new hires have a smooth transition, enhancing overall satisfaction and productivity.
Compliance and Risk Metrics That Apply to Small Businesses
Compliance is where outsourcing genuinely reduces risk, and it’s also where the metrics most often get replaced with reassurance. Three things to measure.
- Filing accuracy, measured by agency notices. Count the tax and regulatory notices your business receives per year. This is the only compliance metric with an external referee, which is what makes it valuable. A provider can characterize its own accuracy however it likes; a state revenue department cannot.
- Audit finding age, not finding count. Ask for the count of open findings and how long each has been open. A provider with six findings all under 30 days is functioning normally. A provider with two findings open for eight months has a remediation problem, and the lower number looks better on a dashboard.
- Documented liability allocation. This is not a metric, it’s the thing that determines whether the metrics matter. Get in writing which party carries liability for a misclassification, a missed filing, or a wage and hour error. Under a PEO co-employment arrangement, some liability genuinely transfers. Under an administrative services agreement, usually far less does than clients assume. Many small businesses discover this distinction during a dispute rather than during procurement.
On data security specifically, the useful questions are about attestation and access control, not architecture. Ask whether the provider holds a current SOC 2 Type II report and will share it, how role-based access is enforced and reviewed, whether MFA is mandatory for administrative accounts, and what the breach notification timeline is contractually. These are the same evaluation fundamentals that apply to any vendor holding sensitive business data, and the red flags to watch for when choosing an IT support provider translate almost directly to HR vendor diligence.
Running the Quarterly Review
Metrics that nobody reviews on a schedule are decoration. A one-hour quarterly session with a fixed structure is enough:
- Twelve metrics against the previous quarter and against your pre-outsourcing baseline. Trend beats snapshot every time.
- Every SLA breach, individually. Not the aggregate percentage. Ask what happened in each case and what changed as a result.
- Every open compliance finding and its age.
- One question you cannot answer from the dashboard. Something like: what’s the single thing most likely to go wrong next quarter? The answer tells you more about the relationship than any metric.
Two governance items to establish at contract signing rather than at the first bad quarter. Metric definitions in writing, so “resolved” and “on time” mean the same thing in month 18 as in month one. And a named accountable contact, because the difference between a provider that fixes problems and one that explains them usually comes down to whether one person owns the relationship. These are the same accountability structures that determine whether any outsourced function works, and they’re covered more broadly in this guide to choosing a scalable partner for complex work.
HR Outsourcing Metrics FAQs
Compare fully loaded HR cost per employee per year before and after, then add the value of outcomes that changed. The cost side is straightforward arithmetic. The outcome side requires a pre-outsourcing baseline on turnover, time to fill, and payroll accuracy, which is why establishing that baseline before the provider starts matters so much. Without it you’re left comparing your provider’s reporting against your memory, and memory reliably favors whichever story is being told at the time.
Look at three things rather than a single response-time number: the response window by request severity, the resolution window, and what happens when the SLA is missed. That third item is where most small business contracts are empty. An SLA with no remedy is a statement of intent. Also check whether the SLA covers payroll processing specifically, since that’s the function where a miss is most expensive and it’s sometimes carved out.
Pricing varies too widely by model, region, and scope for a single figure to be useful, and any article quoting one is guessing. The more productive approach is to get quotes normalized to per employee per month with an identical scope list, because providers bundle differently and headline prices are rarely comparable. Ask specifically whether benefits brokerage, workers’ compensation, recruiting, and compliance consultation are included or billed separately. Those four are where quotes diverge most.
It depends heavily on industry, and comparing against a cross-industry average will mislead you in either direction. Retail and hospitality run structurally high; professional services run much lower. Pull the figure for your specific industry from Bureau of Labor Statistics JOLTS data rather than from a vendor’s marketing page. More useful than any benchmark: track your own voluntary turnover trend and your 90-day new hire retention, since those move in response to things you can actually change.
eNPS. It asks how likely an employee is to recommend the organization as a place to work, which is a different question from customer NPS and sits on a different benchmark range. Treat the trend as the signal rather than the absolute score, and pair it with something qualitative, because a single number tells you sentiment moved without telling you why.
Three signals, in order of reliability. Rising rework rate, meaning more transactions need correction. Lengthening resolution times on routine requests, which usually means your account has been reassigned or the team is under-resourced. And employees routing around the provider by bringing HR questions to you directly, which is the clearest signal of all and appears in no dashboard.
Typically when headcount reaches a point where per-employee pricing exceeds the fully loaded cost of an in-house HR generalist, which for most businesses lands somewhere between 40 and 80 employees depending on complexity and region. The trigger to watch is your own trend line, not a headcount number from an article. Run the per-employee comparison annually and the crossover will announce itself.
Payroll accuracy and on-time rate above everything else, because those are the failures employees notice immediately and they set the tone for the entire relationship. Then rework rate, since early rework predicts long-term cost better than any other early indicator. Talent metrics like time to fill and turnover need two to three quarters before they mean anything, so resist judging the arrangement on them early.
Making the Metrics Work
The failure mode in HR outsourcing is rarely a bad provider. It’s an arrangement nobody measures, running quietly until something breaks visibly.
Twelve metrics across service delivery, payroll accuracy, talent outcomes, and compliance risk, reviewed quarterly against a baseline you captured before the provider started, is enough to catch problems while they’re still cheap. Establish the definitions in writing at contract signing, keep them fixed, and pay closer attention to rework rate and finding age than to the headline percentages, because those two move first.
And treat the industry’s own research as directional rather than predictive. Outsourcing HR can meaningfully improve administrative reliability and benefits access. It will not fix a retention problem that originates in compensation or management, and no provider dashboard is going to tell you which one you have.
