eNPS done right — 3 questions, monthly cadence, anonymous by default

eNPS — employee Net Promoter Score — is the survey question "how likely are you to recommend [company] as a place to work, on a scale of 0–10?" with the famous P-D = N math on top. It's the easiest engagement-measurement habit to start. It's also one of the easiest to mess up.

This piece is the short version of what we've seen work: 3 questions, monthly cadence, anonymous by default, and what to actually do with the results.

The math, briefly

Each response is bucketed:

• 9–10: Promoter
• 7–8: Passive
• 0–6: Detractor

eNPS = % Promoters − % Detractors. Range: −100 to +100. Tech-industry "good" is around +20 to +40; under 0 is a warning sign; +60 is suspiciously high (people aren't being honest).

The score is less interesting than its trend over time. Absolute numbers are noisy; month-over-month deltas are the signal.

The 3 questions

A serious eNPS survey is exactly 3 questions. Not 5, not 12, not 23.

Q1 — the score: "How likely are you to recommend [company] as a place to work, on a scale of 0–10?"

This is the eNPS question. It must be 0–10 (not 1–5, not "agree/disagree"). Two-decade industry standard; don't deviate.

Q2 — the why: "What's the main reason for your score?" (free text, optional)

This is where the actual signal lives. The score tells you the temperature; this tells you what's wrong (or what's right). A team with eNPS +40 and recurring text answers about "great team, frustrated with leadership decisions" is a different situation than a team with eNPS +40 and text answers about "love the work, great pay."

Q3 — the one thing: "If you could change one thing about working here, what would it be?" (free text, optional)

This is the actionability question. Q2 explains the current score; Q3 tells you what would move it. Different question, different answers. Don't combine them.

What to NOT add

A standard pattern: HR launches eNPS, then someone says "since we have everyone here, let's ask a few more things." Three months later the survey is 12 questions and response rate has dropped to 30%.

Specifically don't add:

• Manager-rating questions ("how would you rate your direct manager"). Should be a separate, less-frequent survey.
• Specific-program questions ("how do you feel about the new wellness initiative"). Belongs in a program-specific pulse, not eNPS.
• Demographics ("years of tenure", "team", etc.). Already in the HRIS. Don't ask people to re-classify themselves.
• Multi-select preference questions. Different exercise.

The 3-question rule is its own discipline. Keep adding things and the response rate decay will hide everything you're trying to learn.

Cadence

The two failure modes:

Too frequent (weekly): survey fatigue. By month 3, people answer reflexively without thinking. The signal goes flat.

Too infrequent (quarterly or annually): no trend. The next data point is too far away to spot patterns. By the time you notice the score dropped, the cause was 4 months ago.

The sweet spot: monthly. Often enough to see trends, rare enough that people think about their answer. Make it the same window every month (first Monday → answer by Friday). Predictability matters; surprise surveys produce worse answers.

Anonymity contract

Anonymous-by-default is non-negotiable. People won't write "the CEO is making bad decisions" with their name attached, no matter how psychologically safe you've made the culture.

But "anonymous" needs to mean something specific:

• Score is per-team aggregate, not per-person. The system shows "Engineering team average: 7.3, response rate: 18/22"; never "Igor scored a 4". HR cannot see individual scores even with admin access.
• Free-text answers are forwarded WITHOUT employee identification, but with team-level tagging ("from someone in Engineering"). So a leader can see patterns by team without being able to point at a person.
• Small-team protection. If a team has <5 respondents in a given month, the score isn't shown for that team — too easy to triangulate. The respondents roll up into the parent department.

Without this design, "anonymous" is a polite fiction and your survey is broken on cycle 2.

What to do with the score

The single most common mistake: launch eNPS, get a number, share it on a slide, do nothing. Score next month is unchanged or worse. Repeat for 4 months. Stop running it.

The pattern that produces results:

Month 1: Run the survey. Read every text answer. (Yes, every one. Even at 200 people, that's a 90-minute task once a month and the most important 90 minutes you'll spend on people-ops.)

Month 1, after reading: Identify the top 2 themes in the text answers. Not 7. Two. These are the things multiple people independently brought up.

Month 1, action: Pick ONE of the two themes you can actually do something about in the next month. The other one parks for now.

Month 2: Communicate to the team. Single Slack post: "Last month's eNPS was X. The two recurring themes in your answers were A and B. We're acting on A this month by doing [specific thing]. B is on the list but not now — here's why."

Month 2: Run the survey again. Read every answer. Look for: did A get mentioned less this month? Did the people who mentioned A specifically say they noticed your action?

Month 3: Communicate the trend. Pick the next theme.

The compound here matters. After 12 months of this, you've made 12 specific changes responsive to actual employee feedback, you've communicated each one explicitly tying it to the survey, and the team has learned that the survey produces visible results. The response rate stays high because answering feels worth doing.

If you skip the "communicate what you did" step, the survey rots within 6 months no matter how good your actions are. The communication is half the value.

A note on Pulse Surveys vs eNPS

The Pulse Survey category broadly includes eNPS but also custom-question surveys (mNPS for manager NPS, project-specific NPS, training-feedback surveys, etc.).

A reasonable cadence stack:

• Monthly eNPS — the 3 questions above. Whole company.
• Quarterly mNPS — "how would you rate your direct manager?" Same 0–10 scale. Per-person scores aggregated up to manager level (and only shown when the manager has 5+ direct reports, same small-team protection).
• Ad-hoc custom pulses — when a specific initiative ends (training program, return-to-office change, etc.), a short survey scoped to that thing.

The temptation to bundle everything into one quarterly mega-survey is exactly what eNPS is meant to fix. Keep them separate, keep them small, keep the cadence honest.

How DTPulse does this

We ship eNPS as a built-in feature with the monthly default cadence. 3-question format (Q1 enforced 0–10, Q2 + Q3 free text optional). Per-tenant configurable cadence (weekly, biweekly, monthly, quarterly) with the warnings mentioned above.

Anonymity model: scores aggregated per team, individual answers tagged with team only, small-team rollup at <5 respondents. Admin sees aggregate trends and text answers (without names); reviewees never see who answered what. The mNPS variant works similarly with per-manager aggregation.

The result-sharing step is left to the team — we don't auto-publish to Slack because the "tell people what you're acting on" framing is more important than the raw number, and that needs a human author.

Related reading

• Building a 360-degree review process people don't dread
• What is an employee portal — and when does a company really need one