The Frequency

I evaluate the benefit, not the employee. I ask questions about the thing: when is it available, how many steps it takes to reach, what it costs someone to use it. None of that requires knowing who walked through the door.

When I go and sign off on invoices for detox and residential facilities, the vendors start sending reports, and some of them send more than I should ever see. So I tell our broker to stop sending one of them. If a number could be traced back to a single nurse on a single unit, it doesn’t end up on my desk, because when I see it, I can’t unsee it in a compensation conversation.

I pulled appointment times in aggregate for one benefit, and almost all of it landed between nine and five. We staff around the clock. Half our people were being offered something that was functionally closed when they were awake. The practice that changed my mind was looking at the clock instead of the roster, and I was ready to call the benefit weak. The weakness wasn’t in the benefit. It was the hours.

We pushed the vendor on evening and weekend access before we touched the budget line. The value argument of low usage is likely just a scheduling issue.

Employees should be able to criticize a benefit without revealing circumstances. Build evaluation around anonymous cohorts, separating feedback from identity and performance records. I prefer facilitated listening sessions that report themes after enough participants contribute, especially for benefits involving mental health, caregiving, or debt.

The metric that changed a renewal decision was time-to-confidence, the number of days between learning about a benefit and feeling comfortable enough to try it. It revealed a failure mode, a benefit can be available yet psychologically inaccessible. A long interval pointed to unclear privacy boundaries or cumbersome enrollment, not weak demand. Fixing those conditions preserved dignity and was more defensible than replacing a benefit on thin utilization data.

I favor measuring the confidence people have in using a benefit, rather than measuring every click, appointment, or claim. A short anonymous pulse can ask whether employees know when the benefit is relevant, where to begin, and whether they could use it without involving a supervisor. Those answers reveal accessibility without turning personal decisions into workplace data.

That approach changed the view of a financial education resource that had low visible engagement. Anonymous results showed strong interest but weak confidence about how to start, especially among newer employees. The problem was not the benefit itself. It was the introduction and language surrounding it. Replacing dense announcements with plain examples and private access instructions improved perceived usefulness. Confidence to use a benefit can be a more honest metric than usage volume alone.

Privacy is protected by refusing ethically unusable questions. Benefit reviews should measure access, clarity, and relevance, not diagnoses, family circumstances, or personal usage. An annual anonymous survey, with results reported only in population-level themes, gives leaders enough evidence to make choices without turning care into a data trail.

The metric that changed an earlier decision was the percentage of employees who could accurately explain a benefit. An eldercare resource had enrollment, but few knew where to begin urgently. The issue was not value, it was discoverability under pressure. Clearer entry points and scenario-based education improved confidence, proving that a benefit can be useful before anyone uses it.

I evaluate short-term disability benefits by examining population-level claims and absentee trends rather than tracking individual employees, which preserves privacy and avoids making people feel watched. That data-driven review focuses on income disruption and workforce stability instead of individual case details. Reviewing those claims and absentee patterns changed my view on short-term disability: the trends showed it reduced income disruption and led us to keep and strengthen the benefit. Relying on aggregated metrics lets us improve plan design while protecting employee privacy.

The signal I trust most is voluntary usage over time, read alongside how a team’s delivery holds up during its heaviest weeks. Not who logged in, not how long anyone sat at a keyboard. I watch whether people keep choosing a benefit once the novelty wears off, and whether the teams using it are still shipping without the late-night patterns and unused vacation balances that show up before someone quits.

Usage alone misled me. A mental health stipend looked dead on my dashboard because people were quietly uncomfortable claiming it, not because they didn’t need it. So I pair the number with a plain question asked in a scheduled check-in.

What did you use this quarter, what did you want and skip, and why did you skip it. Answers get aggregated, never attributed.

Before any of that, I tell people what is being looked at and why. Aggregate outcomes, not activity. When my team knows the data exists to decide which benefits survive rather than to grade them, the reporting I get back is more honest.

Here is the decision rule I apply. If a benefit shows weak uptake but strong stated demand, I restructure it instead of cutting it. If uptake is low and nobody names it as something they wanted, that budget moves to whatever people keep asking for. Then I tell the team what changed and what drove it.

When observing employee benefits, one should pay attention to the voluntary aspects of adoption and outcomes, avoiding to overly dissect usage patterns which can be perceived as a form of watching. I have learnt this lesson in my own organization, where we have observed that the moment an employee starts to feel that their privacy is compromised for the sake of a benefit, it ceases to become a retention tool and starts working against retention efforts. Our organization has developed a practice of collecting data on the benefit usage on the cohort rather than the individual level and focusing our attention on how the benefits correlate with some organizational metrics such as attrition and unplanned leave rather than taking a closer look at the specific details of the benefit usage.

My encounter with the specific example of benefit evaluation helped me to change my perspective on what makes it effective. The core practice which forced me to reconsider the whole process was our decision to conduct an anonymous survey in the form of eNPS. Initially, I was convinced that the high enrollment into training programs was enough to declare this benefit as effective. I failed to realize that the enrollment was not motivated by how much the employees value the benefit in terms of knowledge gained from the trainings but by the pressure from the management to participate in the program. Hence, whilst the number of participants in the training was high, employee sentiment remained neutral. Only after we started to analyze the opinions of the workers about the benefits and how likely they were to recommend the benefit to other employees, we learned that the benefits that had the lowest number of participants turned out to be the most appreciated by the employees in our delivery team.