Data in HR (Part 1)

Why NPS is Not the Right Metric

4 minute read

Though a very common tool used in popular surveys like Best Places to Work,  Net Promoter Score (NPS) is not the metric you should rely on to determine employee engagement.

To calculate NPS, companies survey employees (or customers) asking on a scale of 0 to 10, how likely they are to recommend the company to a friend. Once data is gathered, a single number is calculated. 

The NPS was developed by Fred Reichl in 1993 while he worked at Bain & Company. Both his company’s prominence and Reichl’s three books on the topic account for much of its popularity. He continues to speak on NPS and loyalty today. 

Utilizing metrics to understand the employee experience is incredibly useful. NPS is just not the tool I think you should be using. I will go more in depth about why NPS is not effective, but also stay tuned for Part 2 where I discuss what metrics to look at to improve your employee experience.

Why shouldn’t you use NPS?

1. The NPS calculates an arbitrary number.

The way in which the final NPS is calculated can easily disguise major issues or successes with smaller groups within your organization. It does this by dismissing anyone who answers with what it defines as passive scores, and then setting the "detractors" and "promoters" as diametric opposites that can cancel each other out in the final score.

We calculate NPS by counting everyone who answers the survey and we total the numbers as follows:

Detractors (0-6) = D

Passives (7-8), these are thrown out and not used in the formula 

Promoters (9-10) = P

Total Respondents = T

Then, subtract the percentage of Detractors from the percentage of Promoters and multiply by 100 to determine your overall Net Promoter Score.

Net Promoter Score = [(P/T) - (D/T)] * 100

Here are two examples, of 100 respondents who gave different answers but when calculated we get the same score. Both of these graphs show NPS of 0.

Company 1 

(25/100 - 25/100)*100 = 0

Company 2

(40/100 - 40/100)*100 = 0

If we simply looked at the final score, we would think these companies had equally satisfied employees, however Company 1 has many fewer people at the low part of the answers (0 - 5) than Company 2. Company 2 has a lot of people who answered 0, making me think there is a group of really unhappy employees out there that need to be immediately addressed.

2. Societal pressures can impact survey results.

In the 2016 Presidential Election most polls showed Hilary Clinton beating Donald Trump by 1 to 7 points. As we saw after the election, what the polls showed and how people behaved were very different. There are many theories as to why the polls were wrong, but one common one is the nonresponse bias. This is where people don’t respond or delay their response if they think their response is not desirable to others. We know that what people say and what they do are not always aligned and this seems to have played a part in this example. This is certainly a risk we take with employee surveys too, especially if employees feel a lack of trust with the survey’s level of confidentiality.

3. The numbers are not well defined.

Besides telling us that 0 is ‘not likely’ and 10 is ‘very likely’, most NPS surveys do not define the numbers further than that, and the difference of one number can drastically change the final score.

Before I had an understanding of how the NPS was scored, I have certainly put a 7 or 8 not knowing my response wouldn’t count and thinking that a lot of the time I would certainly recommend the company to a friend. I also have put 6 before, not considering myself a detractor at all. 

The reality is that everyone views these numbers differently and unless they are better defined, it can’t give us an accurate idea of how people are really feeling.

4. You should be measuring employee engagement not NPS.

An employee recommending your company to a friend is not correlated to employee engagement or retention. Just because they recommend that their friend works for you does not make them a high producer, highly engaged themselves, or mean they plan to stay with the company. Those are the things you should be measuring.

What should we do?

If you do want to use this metric I recommend making some changes to the way it’s assessed. First, define the numbers from 0 to 10 further or reduce the number of options (1 to 5 can be good). Second, look at the answers on a graph as shown above, versus putting all the weight into a calculated score. Lastly, NPS should never be used alone. Using several metrics to get a holistic look at your employee experience is extremely important. 

Next week I will share some uncommon metrics that I think are great for assessing employee engagement and helping to make improvements where needed.