Net Promoter Score: One Metric to Rule Them All or Just Fuzzy Math?

NPS can be as polarizing as country music. Both lovers and haters are all too willing to make their case to anyone within earshot, equally loud and proud in sharing what makes NPS either the one metric to rule them all or the source of every evil inflicted upon mankind.

Take, for example, UX luminary Jared Spool, who has sustained a long-running campaign to discredit NPS. To be fair, he makes perfectly well-reasoned arguments that accurately call out some of the limitations of the metric — that it’s opaque, ambiguous, and can be gamed — but his sentence doesn’t fit the crime. Spool tells us to run! run fast and far! treating NPS like some sort of scourge or airborne infection rather than the imperfect metric it is.

The reality is that kicking NPS to the curb would be, at best, a pyrrhic victory, one that satisfies some intellectual desire for statistical purity, while neglecting what NPS does pretty darn well.

Why NPS is Worth Measuring

It’s simple

The first and perhaps most obvious advantage of NPS is that it’s super simple. One single question that, in the opinions of most, provides insight into customer sentiment by measuring their stated willingness to advocate. In his seminal HBR article on NPS, Fred Reichheld wrote that, “When customers act as references … they put their own reputations on the line. And they will risk their reputations only if they feel intense loyalty.” With one simple question, NPS measures the degree to which customers are willing to put their reputations on the line for your product.

It’s comparable

The fact that NPS is simple means that it’s widely adopted. Roughly two-thirds of the Fortune 1000 use NPS and, according to Pendo’s State of Product Leadership survey, NPS is second only to revenue as the north star KPI that product leaders follow. This means that there’s a ton of data available for benchmarking your NPS performance against other companies.

It’s extensible

Perhaps the most frequent criticism of NPS is the fact that it’s not particularly actionable. Knowing how customers feel isn’t nearly as useful as knowing why they feel that way. But this limitation is hardly unique to NPS. Virtually every type of survey question depends upon further elaboration and illumination. For example, an NPS best practice is to add a follow-on question (known as a verbatim) that asks the critical question: why? The NPS question doesn’t have to stand on its own. It’s one of a constellation (or a small collection, anyway) of metrics that you ought to track. When you combine NPS data with quantitative data like product usage or transaction data, you begin to shine light on how your product creates or destroys user happiness and how that sentiment impacts revenue.

The need to dig deeper and look further isn’t unique to this particular methodology and shouldn’t mean that NPS data isn’t worth collecting in the first place.

What Spool Gets Right

Let me restate that Jared Spool isn’t all wrong about NPS. It’s an imperfect metric. He correctly points out that some of the methodology design choices are, in places, curious.

As a refresher, NPS is based on an 11-point scale, where 10 equals “extremely likely,” and zero means the exact opposite.

NPS Scale

Instead of calculating an average of the scores, NPS segments zero through six as detractors, sevens and eights as passives and nines and tens as promoters. The score is calculated by subtracting the percentage of promoters from the percentage of detractors.

The consequence of this design choice is opacity at the low end.

For example, a score of zero yields an NPS score of -100, the worst possible outcome. Because zero and six are assigned the same value, a score of six yields the exact same NPS score. What this means is that, while NPS can effectively measure the progression of good to great and great to exceptional, it’s not useful for measuring the progression of terrible to adequate.

Spool’s concern, I believe, is that as a UX designer, this could potentially understate earlier progress made in the direction of great. This is both a fair concern and evidence that Spool is perhaps misunderstanding where NPS fits in a broader CX context.

What Spool Gets Wrong

Spool’s harsh critique seems to derive from a fairly black and white view about NPS, assuming that it’s presented as the only metric that matters. This simplistic notion is only reinforced by Fred Reichheld himself. In 2006, the NPS inventor published a best selling book entitled The Ultimate Question, which popularized NPS with this somewhat oversold promise.

By suggesting that NPS is the one metric to rule them all, Reichheld is only shining light on its limitations–and inviting the harsh scrutiny of Spool and others.

The best way to think about NPS is that it’s both necessary and insufficient.

Particularly for UXers like Spool. Simply asking the NPS question isn’t enough. Instead, they need to add richness to NPS data by pairing it with behavioral analytics that connect sentiment to usage, and asking the all-important why?

Just as importantly, they need to think of NPS as one of three questions that you ask at various points on the customer journey:

  1. Use CSAT to ask “How would you rate your overall satisfaction with X?”
  2. Use Customer Effort Score (CES) to ask “how easy was it to do X?”
  3. Use NPS to ask “how willing would you be to refer X to a friend?”

Taken together, these questions tell you a heckuva lot about your customer.

Why NPS Should Matter to Product People

Spool points out that NPS scores are a bit of a black box at the low end. If a user is a six or less, they’re among an indistinguishable mass of detractors. The truth is that this is the death zone for digital products. Today, product is the new marketing. Good enough isn’t good enough.

NPS is unapologetically a measure of product love. It helps you understand the extent to which your users are raving fans.

And what if they’re not? Then you need to understand why–which leads you to a follow-on verbatim, an overlay of behavioral data, and the application of additional survey questions to better understand the root causes of positive and negative sentiment.

Does this mean that you shouldn’t measure product love in the first place? Hardly. It simply means that NPS is just the beginning–the starting point, and not the final destination.

(Interested in measuring NPS in your product? My company, Pendo, just launched Pendo Vox, a free tool for collecting and analyzing NPS data in any app. It’s free forever. Get started here.)

About the Author

Jake Sorofman is CMO of Pendo. Most recently, he spent five years as a VP and Chief of Research at Gartner, Inc., before returning to his startup roots. His writing has appeared in Harvard Business Review, Forbes, and other business and industry publications.