Employee Net Promoter Score

eNPS adapts the customer net promoter score to employees. The question is how likely you are to recommend this organisation as a place to work, answered from nought to ten.

The score is the percentage of promoters minus the percentage of detractors, producing a figure between minus one hundred and plus one hundred. A score of zero means promoters and detractors are equally numerous, which is a considerably worse position than the number suggests.

Two features of this calculation cause most of the misreading. The middle is thrown away, so a large group of reasonably positive people contributes nothing. And the detractor band is very wide: a six, which most respondents intend as moderately positive, is counted the same as a nought.

Neither is a defect exactly. The bands were chosen to be demanding, on the reasoning that only genuine enthusiasm counts as advocacy. But an organisation reporting the number without understanding the bands will misinterpret both its level and its movements.

A single number carries a single piece of information, and eNPS is often asked to carry more.

It cannot say why. Two organisations with identical scores may have entirely different problems, and the same organisation can hold a stable score while the reasons behind it change completely.

It cannot distinguish between a workforce that is uniformly lukewarm and one that is sharply divided. A score built from many passives and a score built from equal promoters and detractors can be identical and describe very different situations, and the second is considerably more urgent.

It also moves for reasons unconnected to how the organisation is run. A redundancy round, a change of ownership, a bad quarter in the industry or a widely reported news story about the sector all move the score.

The standard remedy is a free-text follow-up asking the reason for the rating, and this is where nearly all of the usable information lives. Organisations that report the number and never read the comments have collected the least useful part of what they gathered.

Despite the limitations, the measure has real advantages and they are mostly about response.

One question gets answered. Response rates for a single item are substantially higher than for a long instrument, and they hold up under repetition, so the measure can be run frequently without exhausting people. That makes it suitable for tracking change through a period of disruption, where an annual survey would arrive far too late.

It is also intuitive to report. Leadership teams understand a single trending figure, and a measure that gets attention imperfectly is more useful than a sophisticated one that gets ignored.

And the question itself is a reasonable proxy. Recommending an employer to someone you know carries a small personal cost if the recommendation turns out badly, so a positive answer implies a degree of confidence that a satisfaction rating does not.

The sensible use is therefore as a frequent, lightweight trend indicator sitting alongside a fuller instrument run less often, rather than as a replacement for one.

• Track the trend within your own organisation and treat external comparisons with suspicion, since scores depend on how the question was asked, of whom, and when.
• Read it by team, site and tenure. As with engagement, the aggregate hides the distribution that matters.
• Report the three bands alongside the score, because promoters at forty per cent with detractors at forty per cent is a different situation from a large passive middle, and the single number cannot distinguish them.
• Always pair it with the free-text reason, and read the text rather than only counting themes.
• Watch what happened in the period. A drop that coincides with a restructure is explained; one that does not is the interesting case.

It is also worth being careful about anonymity. Because the instrument is short and often run frequently, it tends to be deployed at team level, and a team small enough to identify individuals from their responses will produce cautious answers regardless of what the introduction promises.

Where teams are small, aggregating to a larger unit gives a more honest reading than a granular one nobody trusts.

Any single measure reported upwards eventually becomes a target, and eNPS is unusually easy to influence without changing anything real.

Timing the survey after a positive announcement, prefacing it with a message about how well the organisation is doing, running it immediately following an increment, or having managers encourage their teams to rate generously all move the score. None of them moves the underlying position.

The signature of this is a score that improves while attrition, absence and the free-text comments do not. Where those diverge, the score is the one to distrust.

The defence is to keep the collection consistent, at the same points in a cycle and with the same wording, and to hold the score against measures that are harder to influence. Attrition, particularly regretted attrition, and unplanned absence both serve as a check.

Where a manager's team score is used in their own assessment, the incentive to manage the number rather than the team is direct, and it is worth thinking hard before creating it.