The Employee Net Promoter Score (eNPS) is derived from the Net Promoter Score (NPS) popularized in customer and market research. In simple terms it refers to a simple score derived from a single question asking people how likely they'd be to recommend working at their organization. The question itself is a good one when asked in the right way and we use a 5-point variant of this question in our standard Employee Engagement Index. We will soon be releasing an eNPS feature that will calculate an eNPS score using the traditional NPS scoring method (contact your coach if you want to learn more about that). For now, here is a short summary of our thinking and research on eNPS.
1. The customer oriented 0-10 NPS scale isn't necessarily ideal practice.
The customer oriented scale asks people to rate how likely they are to recommend on a scale from 0 to 10 (it's an 11 point scale). This was a method that was quite popular for telephone interviews where reading out scale labels to people can be time consuming. With so many points it is usually unfeasible to label all the scale points and it is also much less user friendly especially on a mobile device. Furthermore, a lot of evidence suggests that labelling each point on a scale is one of the main things that improves the reliability and validity of scales.
2. The 0-10 eNPS scale seems to measure the same thing as a 5 point scale
We conducted a case study with a customer where we had ~500 people use a 0-10 eNPS question as well as our standard 5-point question. The results were close to perfectly the same with a correlation of .91 and the 0-10 and 5-point eNPS scores were both around 37 (decimals apart at 36.6 and 36.7 in this case). Here's what the data looks like.
Smoothed density scatterplot of eNPS scores using 5-point and 0-10 (11-point) scales. The regression line shows a near perfect correlation (r=.91). N.B. Darker zones represent larger numbers of respondents in the same spot on the chart.
3. The 0-10 eNPS scale or any other single question is never as reliable an outcome measure as a multi-question index
It is a commonly known statistical recommendation that when it comes to self-report scale metrics for an important outcome we are far better off using a multi-question approach (we combine 4-5 questions into a single index for Engagement for example). Even within the customer and market research studies have shown that using multiple questions is more reliable and accurate than using a singular NPS type metric or question. Our own data observations tell us that single item outcome measures result in weaker statistical outcomes and reliability.
Reliability levels (Cronbach's Alpha) shown as the number of questions in an index is reduced. Reliability below .70 are considered questionable. Below one it is assumed that reliability has dropped further although it is not calculable.
We support asking the eNPS type question in our 5-point format but using a simple percentage favorable. However, get in touch if you’re still convinced you need eNPS (e.g. for external or historical benchmarking), as we may be able to help you navigate a transition.