Every year, you send out that employee benefits survey. You get back a spreadsheet full of neat averages and satisfaction scores. You nod, file it away, and move on.
That standard template undercuts your strategy. The data is noisy, and acting on it can expose you to legal risk. Four failures stand out, each one with a fix.
1. The Average Is a Lie
Your survey asks: "How satisfied are you with your healthcare costs?" Then you average the answers. You have created a number that means nothing.
Consider two employees:
- A healthy 24-year-old on a low-deductible PPO with zero claims says "5 - Very Satisfied."
- A 58-year-old diabetic on an HDHP with a costly specialty drug says "1 - Very Dissatisfied."
Average them? You get a solid 3. Your leadership sees that and says, "Healthcare costs are fine, no changes needed." Meanwhile, your highest-cost population is drowning.
The fix: Stop averaging across everyone. Segment your respondents by plan type, age, chronic condition status, and claims tier before you calculate anything. The overall average is noise. The cohort breakdown is where the real story lives.
2. You're Asking Questions the System Already Answered
A classic template question: "Did you find Open Enrollment easy to navigate?"
Your employees guess. Some say yes, some say no. You get a 7/10 and call it a day.
But your Benefits Administration system recorded different numbers:
- 40% of employees abandoned their enrollment cart mid-way.
- Average time-on-task was 14 minutes for a 3-minute process.
- The support line lit up with the exact same question about dependent verification.
You're paying employees to guess at something your platform has already measured. It's a waste of everyone's time.
Worse yet: "Do you understand your HSA?" Your survey says 85% understand. But your system shows 60% never funded their HSA, and 15% took non-qualified distributions. The survey says everything is fine. The system screams it's broken.
The fix: Replace generic satisfaction questions with specific, system-informed prompts. Instead of "Was Open Enrollment easy?" try: "We noticed you started enrollment but didn't finish. What happened?" That question is a feedback loop driven by actual behavior.
3. Asking About Health Is Itself Regulated
Segmenting by chronic condition status sounds like clean advice until you remember how an employer learns an employee has a chronic condition. If the survey asks, the employer is collecting health information, and that step has rules attached.
The Americans with Disabilities Act restricts employer disability-related inquiries and medical exams. An employer can ask about health status as part of a voluntary health program, but a mandatory survey question about chronic conditions sits on regulated ground. The Genetic Information Nondiscrimination Act requires prior, written, knowing, and voluntary authorization when questions touch family medical history. In AARP v. EEOC, a federal court vacated the 2016 incentive limits, which unsettled the standard for what incentive keeps a health questionnaire voluntary. Replacement rules are still not finalized.
The survey is the wrong tool for this data anyway. Your claims file already contains diagnosis codes, tier assignments, and chronic condition flags. Pull segmentation from there. Claims data does not require an employee to self-identify, does not add an authorization step, and reflects the treatment that happened rather than what an employee is willing to type into a form.
The fix: If you are segmenting by health status, use the claims file you already own and keep health questions out of the survey. If a survey must touch health information, treat it as a regulated employee health inquiry: make it voluntary, get the right authorizations, and have counsel review the questions before they ship.
4. Changing a Benefit Is Not a Fiduciary Act
Under ERISA Section 404(a)(1)(B), plan fiduciaries must act "with the care, skill, prudence, and diligence... that a prudent man acting in a like capacity" would use. That duty is procedural: a fiduciary builds a documented, defensible process and follows it each time.
But changing a benefit is not a fiduciary decision. Courts and the Department of Labor treat plan design as a settlor function, an employer act outside ERISA's fiduciary rules. Dropping a fertility benefit or changing a formulary is plan design. The prudence duty does not reach it.
Prudence does reach the fiduciary side: picking a vendor, evaluating a service provider you delegated authority to, deciding whether to keep a benefits administrator. If survey data informs those moves, keep a written record of what you relied on and why.
The legal risk from a survey-driven design change mostly lives in discrimination law. A fertility coverage cut looks like sex and pregnancy discrimination under Title VII. Restricting mental health benefits can implicate the Mental Health Parity and Addiction Equity Act. Changing drug tiers in a way that lands on employees with disabilities can draw an ADA challenge. A plaintiff's lawyer focuses on whether the change produced a disparate impact the employer cannot justify.
The fix: Keep a written survey protocol and decision log for any survey that informs a change: who was invited, who responded, how results were weighted, and whether the sample skews toward the loudest group. Run any benefit cut or formulary change past employment counsel first. If the survey feeds a genuine fiduciary decision, that same log is your prudence record.
The Future: Surveys That Work
Treat the survey as a validation layer on top of the data you already own.
- Replace satisfaction with friction. Let your Benefits Administration system flag pain points, such as an employee who searched for an in-network therapist three times without success. Then ask why or what else.
- Stratify everything. Pull lifecycle data: age, tenure, salary, dependents, plan enrollment. Report by cohort.
- Audit every question. Before it goes live, ask: "Can I verify this answer with system data?" If not, the question is measuring opinion.
The Bottom Line
The generic benefits survey template is a relic from a time when HR had no data. Today, behavioral data sits in your HRIS and benefits platforms. If your survey isn't talking to those systems, you aren't surveying employees. You're polluting your strategy with noise.
Stop running the script. Start running the system.
This article is for general information only and is not legal advice. Employers should consult their own advisors.