It’s hard to describe the impact the Net Promoter Score had when it hit the Harvard Business Review in 2003.

The business world went nuts. Because in its most boiled-down form it contained only one question – “How likely are you to recommend [product or service] to a friend?” – the NPS could fit practically anywhere: on a postcard, at the end of a business transaction, in an email, as a popup over your favorite website, in a coat pocket, or on the end of a pin.

All of a sudden people who had never gone near a survey for fear of being sucked into a flaming vortex lined with Pearson’s coefficients were sitting around at parties casually reciting their NPS numbers like they were the scores of their kids’ soccer games.

It’s gotten so that everything in the universe has become NPS-able: parties (“On a scale of 1 to 10, how likely are you to recommend Bruh and Buddha’s kegger to a friend?”), binge eating (“On a scale of 1 to 10, how likely are you to recommend to a friend eating an entire pallet of Ding Dongs and washing it down with a five-gallon can of Vernors?”), and even mountain-bike crashes (“On a scale of 1 to 10, how likely are you to recommend to a friend getting a totally gnarly road rash by losing your back end halfway down the Devil’s Vermiform Appendix?”).

Well, not me. I call BS on the NPS.

To me, the NPS as it’s most commonly used – one question, lobbed at you seemingly at random – is an absolutely awful measure of customer behavior.

Let Me Explain ...

You’re a business, and you’re looking for an end result.

Photo credit: Dmitri Popov via Unsplash.

Your goal is to maximize the number of end results. It doesn’t really matter if those end results come from one customer or millions. You just want the highest possible number.

So far, so good. Now, what’s the customer behavior that’s going to get you that end result?

The most direct behavior is a customer saying, “I’m going to do this end result,” and then actually doing it.

There are many indirect behavior chains that could get you that end result, including the one you want, per the NPS: Potential customer asks current customer what they think of your end result, current customer tells them, potential customer acts, and the end result is produced.

But these chains are indirect. They rely on multiple behaviors firing in their proper sequence, and as America's Funniest Home Videos teaches us, when does that happen?

Think of all the areas where that desired behavior chain could fall apart:

The potential customer doesn’t ask anyone before making a decision, but reads online reviews, asks a salesperson, looks at the package, or engages in other perfectly reasonable types of consumer behavior;
The potential customer doesn’t know you’re an expert on this end result and instead asks someone who thinks the earth is actually a very large Ritz cracker;
The potential customer asks lots of people what they think of the end result, and your input is just one of many data points;
They ask you but forgot what you said, because it wasn’t that important anyway; or
They ask you and do the opposite, because them.

And that’s the consumer behavior the NPS purports to measure.

I understand that word-of-mouth is the golden oasis of marketing, and the NPS is seen as the path through the desert, but it’s a mirage.

“The NPS measures the potential willingness to perform a behavior that will only produce the desired end result in one scenario out of many.”

Why? Because the NPS measures the potential willingness to perform a behavior that will only produce the desired end result in one scenario out of many.

And we haven’t even gotten into the question of what a six versus a seven really means on a 10-point scale that measures the possibility that you might recommend a product or service you’ll probably never be asked about.

Where It Fails

Adobe Photoshop is the perfect example for me of where the NPS fails.

I am an incredibly bad Photoshopper. I’m lucky if I can remove the background from clip art and not have half the actual art go with it.

However, my ineptitude at Photoshop does not stop Adobe from asking me at least once every other month how likely I am to recommend Photoshop to a friend.

This is an absolute fairies-and-unicorns question, because:

I am horrible at Photoshop;
I know at most one-tenth of 1 percent of Photoshop's capabilities;
People who know me know I am horrible at Photoshop, and thus would never ask me what I think of the program; and
If on the off chance someone would ask me how I feel about Photoshop, I would have to say, “You do know that chickens are better at Photoshop than I am,” in which case they’d ignore everything I say.

I ran this logic past my colleague Deb, and she went along with it, but then she asked: “Okay, but what question would you ask if you only had one question?”

I thought a moment, then went back to my desired end result and answered, “I’d ask the customer this: 'Would you buy the good or service again?'”

“Given a measurement of an indirect path to a referral sale and a measurement of a direct path to a repeat sale, I’ll take the repeat sale every time.”

Any question asking about future behavior is going to be a bit wobbly, but if I ask someone if they would buy product x again, I can feel at least 40 percent sure they actually will.

Given a measurement of an indirect path to a referral sale and a measurement of a direct path to a repeat sale, I’ll take the repeat sale every time.

And if I ask about repeat purchases, all the good things about the NPS – portability, the ability to track across time – are still in play.

Plus, from that question I can then ask why, and get some data I can really use.

When The NPS Is Useful

Actually, the NPS can be pretty useful if it’s allowed to be more than one question in a box. In fact, I can think of two instances where it’s perfect:

If you have a nerd-geek product and you’re surveying an audience of nerd-geeks. If you’re in charge of marketing for a revolutionary new digital transit, you want to ask the NPS to your customer base because at the end of the day, yeah, surveyors sit around and talk about surveying tools, and recommendations matter.
If you have real influencers. Of course, you want to qualify your influencers, first by asking, “How often do people ask for your recommendation on [product or service]?” and then by asking something like, “How many people on Instagram wait breathlessly for pictures of you making pouty lips?”

Listen, if you have the NPS plugged in everywhere and you feel it’s measuring actual referrals, and you can tie those referrals to your bottom line, more power to you. Rock on.

But let me put this bug in your ear: if you’re looking for something more, why not ask the question you want to ask in the first place?

If You Had Only One Question To Ask, Would You Ask The NPS? Really?