eNPS survey and 1:1s: measuring how people feel, then acting on it

Most companies find out how their people feel at exit interviews, which is too late to do anything with it. An eNPS survey is a cheap, quick way to find out earlier: one question, a follow-up "why", and a number you can track over time. On its own it fixes nothing. Paired with regular one-on-one conversations between managers and their teams, it tells you where to look and gives managers a reason to ask.

What an eNPS survey asks

The employee net promoter score formula comes from customer research. The core question is:

On a scale of 0 to 10, how likely are you to recommend this company as a place to work to a friend?

Then one open question: "What is the main reason for your score?"

Answers fall into three groups:

eNPS = % promoters minus % detractors. The result runs from -100 to +100.

A worked example from a Chennai office

A 600-person services company in Chennai runs its first eNPS survey in July.

• Responses: 420, a 70% response rate.
• Promoters: 160 (38.1%).
• Passives: 150.
• Detractors: 110 (26.2%).
• eNPS: 38.1 - 26.2 = about +12.

The number is less interesting than the breakdown. Operations scored -8; engineering scored +24. The comments from operations kept returning to the same three themes: night-shift transport, unclear shift rosters, and managers who were never available. Those are fixable problems with owners, which is the point of the exercise.

Six months later operations has moved to +3, the response rate is 76% and the transport complaints have nearly gone. The roster comments remain. That is the next quarter's work.

Make the anonymous employee survey actually anonymous

People answer honestly only if they believe nobody can trace their answer. Set the rules before you open the survey and tell people what they are:

• A minimum group size. Never show results for a team smaller than five. A department of three should be shown only as part of a larger group.
• No identifying cuts. Results by "women in finance aged over 40" can point to a single person. Keep cuts broad.
• Shuffle comments. Show written answers in random order, not by time, so nobody can match them to when someone was seen at their desk.
• Separate the reminder list. HR can see who has not answered, to send a reminder, but that list must never be linked to answers.
• No hunting. Managers must not try to work out who wrote a comment. Say so in writing.

Survey answers that can be linked to a person are personal data under the Digital Personal Data Protection Act. Even with an anonymous survey, tell people what is collected, why, who sees the results and how long they are kept.

Pulse survey frequency: how often is too often

Survey fatigue is real, and it is caused less by the number of surveys than by the lack of visible action after them.

A rhythm that works for most companies:

• eNPS twice a year, or quarterly in a fast-changing company.
• Pulse surveys of three to five questions on one topic (a new policy, a move to a new office, a change in shift patterns) when there is something specific to learn.
• No survey in the month after a big announcement or during the appraisal cycle, when answers are distorted.

After each survey, share the result and three actions within a month. "You told us X, we are doing Y, by Z" is what gets people to answer next time.

Reading the comments

The score tells you whether things are getting better or worse. The comments tell you why.

1. Read every comment from detractors, and a sample from passives and promoters.
2. Group them into themes, with a count for each.
3. Pick no more than three themes to act on per cycle.
4. Give each theme an owner and a date.

AI summaries can help with grouping a thousand comments, but someone in HR should still read them. The comments with the most important signals are often the ones that fit no theme.

Where 1:1s come in

A survey tells you that operations is unhappy about rosters. It cannot tell you that Ramesh on the night shift has been doing two jobs since his colleague left. Only a conversation does that.

Regular one-on-ones between each manager and each team member are where survey themes turn into specific fixes. A simple one-on-one meeting agenda:

1. Their topics first. What is on your mind?
2. Work in progress. Blockers, help needed.
3. One survey theme. "People said rosters come out late. How does that affect you?" Never "what score did you give?"
4. Growth. One thing they want to learn or do next.
5. Actions, each with an owner and a date, carried forward to the next 1:1.

Keep shared notes that both sides can see, and let each person keep private notes nobody else can read. Weekly or fortnightly is a good cadence for most teams; monthly is the minimum.

HR's job is to check that 1:1s happen, not to read them. A report showing which managers have not met their people in six weeks, without any notes, is enough to start a conversation. New joiners need 1:1s most of all; they are the subject of our guide to the probation confirmation process.

Mistakes we see

• Announcing the eNPS number to the whole company and nothing else.
• Comparing the score with other companies' published figures instead of your own trend.
• Showing results for teams of two or three.
• Asking managers to "improve their score", which leads to pressure on people to answer 9 or 10.
• Running a survey and then going silent for six months.

A short checklist

• One question, one "why", a clear notice on anonymity and data use.
• A minimum group size of at least five for results.
• Results and three actions shared within a month.
• Themes taken into 1:1s, with actions tracked.
• A check that 1:1s happen, without reading them.
• The same survey, the same way, so the trend means something.

In Vatsin Performance, Engagement surveys include an eNPS type with the 0 to 10 question and a "why", results that appear only once the minimum group size is reached, shuffled comments, and 1:1s with shared and private notes and actions that carry over. For the data side of all this, our guide to the DPDP Act for HR covers notices and retention.