The reasons people leave are rarely the ones they report.

One in four of your employees is caring for someone else.

They're on the phone with a rehab facility during a lunch break. Managing a parent's medication list from a notes app. Using PTO for appointments that aren't theirs.

Others are three weeks into the job and have not yet heard that anyone noticed their work.

They rarely tell you. What you see is absenteeism, presenteeism, and a resignation you didn't see coming.

How the problem materializes

The mechanics of caregiving loss are almost entirely about timing.

Institutions — hospitals, rehab facilities, insurers, benefits administrators, elder-law attorneys — answer the phone between nine and five. So does the caregiver's employer. Every hour of care coordination therefore comes directly out of working time, and the coordination burden is not small: appointments, insurance denials, medication reconciliation, facility selection, financial and legal paperwork, and the slow-motion crisis management of a condition that keeps changing.

The employee absorbs this silently. Disclosing it feels like admitting you're less committed, and in most organizations there is no formal support to disclose it to. So it shows up in your data as something else: unexplained absence, degraded performance from someone who was previously strong, and eventually a departure attributed to "personal reasons."

The population this hits hardest is the most expensive one to lose.

Substance use is the same problem wearing different clothes

The reason is stigma, and stigma is structural rather than attitudinal. Using an employer-sponsored program to disclose a substance-use problem means routing that disclosure through a channel the employer arranged. Whatever the confidentiality policy says, the employee is doing a risk calculation about their job. In a clinical workforce, where licensure and patient safety are involved, that calculation is even less favorable — the perceived downside of disclosure is not embarrassment, it's a career.

So the benefit sits unused by the population that needs it, while the cost — absenteeism, errors, turnover, and in safety-sensitive roles, genuine risk — accrues anyway.

Recognition is the same problem from the other side

Caregiving and recovery support reach people carrying something heavy. Recognition reaches people before they have decided whether this job is worth carrying at all.

The first ninety days decide more than the next five years. A large share of turnover in hourly and deskless healthcare roles happens before the first quarter ends. These people did not leave over pay; most had not been there long enough to have an opinion about it. They left because nothing in those weeks gave them a reason to stay, and a competing offer arrived when leaving cost them nothing.

Clinical staff are recognized constantly by patients and almost never by the organization. That produces a specific and corrosive pattern: people who feel valued by the work and invisible to the employer. That gap is what a competing recruiter is actually selling against, and it does not close with compensation.

Frequency beats magnitude. An annual awards ceremony recognizes a handful of people and, by implication, does not recognize everyone else. What works is specific, timely recognition tied to something the person actually did, early and often.

Why the obvious fixes don't work

A resource directory is not help. Most caregiving benefits are a phone number and a list of vetted providers. That's useful to someone who has time to make calls. The defining characteristic of a caregiver is that they don't.

Awareness campaigns don't overcome stigma. Telling people the EAP is confidential doesn't change their risk calculation, because they already know what the policy says. What changes the calculation is design: support that doesn't require disclosure to the employer at all.

Mental health days don't touch the mechanism. They're a real benefit and worth having. They also don't reduce the coordination burden by a single phone call.

Pizza and swag are not recognition. Generic appreciation reads as generic, and reads worse to a workforce that knows what the same money would have bought in staffing.

What actually moves the needle

Do the work, don't route it. The difference between a directory and a concierge model is whether someone else makes the calls. That single distinction accounts for most of the difference in outcome in this category.

Decouple support from disclosure. Confidential, always-available support that an employee can use without telling anyone gets used by the people who most need it. The employer sees aggregate patterns, enough to see the trend and plan for it, without individual exposure.

Recognize early, often, and without markups. Peer-to-peer recognition and milestone automation, on a reward network that does not take a cut of the award. Markups quietly consume part of many recognition budgets, so the program is smaller than the invoice suggests, underperforms, and gets cut at the first review. Weight it toward the first ninety days, where it does the most.

Judge these programs by who stayed, not by who logged in. Caregiving, recovery and recognition support all pay off in people who did not leave. Evaluate them on engagement percentages and you will undervalue them, because the employee who quietly stayed never shows up in a usage report.

Where to start

BenDNA looks at absence patterns, the shape of PTO use, turnover by tenure band and role, EAP use against benchmark, and current recognition spend. Where EAP use sits far below benchmark, that is not evidence people are fine; it is evidence the channel isn't trusted. Where turnover concentrates in the first ninety days, that is a recognition and onboarding problem, not a pay problem, and it is the cheaper one to fix.