NPS tells you the score moved. Here's how to find out why.
By Nic Chamberlain · Published · Updated
Net Promoter Score has earned its place. It's one question, it's easy to explain to a board, and it gives you a number you can compare quarter to quarter and against competitors. Teams that already track it can keep that measure while investigating the experiences behind it.
But anyone who has sat in the meeting after a two-point dip knows its limit. The score tells you something moved. It doesn't tell you what moved it.
This post is about closing that gap: keeping your NPS score, and hearing the reasons behind it in your customers' own words, from the same conversation.
What NPS can and can't tell you
NPS asks how likely someone is to recommend you on a scale of 0 to 10. You subtract the share of detractors (0–6) from the share of promoters (9–10) and get a single number.
That single number is the strength and the weakness. One customer had a flight cancelled. Another had a great call with support. Both feed into the same score and can cancel each other out. By the time the number reaches a slide, the individual stories that produced it are gone.
Even the score's creator has said as much. Fred Reichheld introduced NPS in Harvard Business Review in 2003. In 2021 he returned to HBR with "Net Promoter 3.0", arguing that the score works best alongside a harder, accounting-based measure of growth from repeat customers and referrals. The number was never meant to carry the whole conversation on its own.
Why the comment box and driver analysis only get you part of the way
Most NPS programmes already try to capture the why, usually in two ways.
The open comment box. "What's the main reason for your score?" helps, but people may write only a few words or skip it. "Price." "Slow delivery." "Great staff." You learn the topic, but not the story behind it, and you can't ask a follow-up question.
Driver analysis. Statistical models split the score into contributing factors such as value or ease. They're useful for prioritising, but each driver is still a label. Knowing that "ease" explains part of the drop doesn't tell you which moment felt hard, or what the customer expected instead.
Both are worth keeping. What they lack is a conversation: the chance to ask "what happened?" and then "why did that matter to you?"
How to get the score and the why in one conversation
Juno runs AI-moderated interviews with real people. You brief it in plain language, it interviews your customers by text or voice in their browser, follows up on what they say, and brings the findings together with every claim linked to the evidence behind it.
Juno interviews aren't only open conversation. They can include structured questions such as rating scales alongside the open discussion. That means the NPS question itself can sit inside the interview, and you still get your score.
1. Brief Juno on what you need to understand
Describe the decision you're facing, the way you'd brief a researcher. For example:
"We track NPS every quarter. Among customers who've been with us under a year it fell four points last time. We need our score as usual, and we need to understand what's driving it."
Juno turns that into research objectives rather than a fixed list of questions, so each interview can follow what matters to that customer.
2. Ask the NPS question the way you always have
Use the standard wording: how likely are you to recommend us to a friend or colleague, from 0 to 10. Calculate NPS as the percentage of promoters minus the percentage of detractors, including passives in the respondent total. The same wording and calculation do not guarantee comparability: sampling, timing, channel and interview context can all change the result. Keep the established tracker as the reference until you have evaluated those differences.
3. Let Juno ask why, straight after the score
This is where it changes. When a customer gives you a 6 and says "delivery was slow", Juno asks what happened and why it mattered to them. The reason is captured in the moment, from the person who gave the score, with the opportunity to clarify a short initial response.
4. Get the number and the reasons together
You can calculate NPS from the structured ratings, alongside findings organised around your objectives, each traced to the customers and quotes behind it. You can see what's pulling detractors down and what's keeping promoters loyal, in their own words. You can question the findings, ask what didn't hold up, and drill into the evidence before you take it to the team.
Juno tells you what customers said and why. What to do about it is still your call, and it's a much easier call when you can see the reasons.
Already running NPS in another tool?
You don't have to move everything at once. Keep your current survey running and send a Juno interview to a sample of recent respondents, mostly detractors and passives, plus some promoters. Knowing what keeps people loyal tells you what not to break. This deliberately selected follow-up sample is for understanding experiences, not for recalculating your overall customer NPS.
If you do move your tracking into Juno, run both side by side for one cycle first. A change in format can shift scores. Evaluate the effects of sampling, timing and channel before comparing with earlier waves; one parallel cycle is a starting point, not proof of equivalence.
When to run interviews alongside NPS
-
After an unexpected movement. A drop, or a rise you can't explain, is the clearest trigger. Interviews run over days, not months, so you can have the reasons before the next reporting cycle.
-
Before a decision that depends on the score. If a pricing change or a service redesign is being justified by NPS, hear from the customers behind the number first.
-
On a regular rhythm. Some teams run a small round every quarter so the why sits next to the what in every report, instead of being commissioned only when something goes wrong.
What this doesn't replace
You keep NPS as your consistent, comparable measure, and Juno doesn't replace your research or insight team. Your team still owns the questions and the decisions. What Juno adds is the conversation the score was never designed to have.
Frequently asked questions
Does Juno replace NPS? No. You keep your NPS score. Juno can ask the NPS question inside the interview and then follow up on the reasons, or run alongside the survey you already have.
Will my score still be comparable? Not automatically. The same question and formula can produce different results when sampling, timing, channel or context changes. Keep the existing tracker while evaluating a parallel interview format.
Can we interview our own customers? Yes. You send them a link to the interview, through whatever channel you already use to reach them. Nothing to download.
How long does it take? Days, not months. Interviews run as customers complete them, so the timing mostly depends on how quickly your customers respond.
How are findings presented? As claims linked to the evidence behind them, down to the quote and the moment in the transcript. You can debrief with Juno and challenge any finding.
Who builds the study? You brief Juno in plain language. It proposes research objectives and checks them for leading wording, and you can rehearse the interview yourself before any customer sees it.
Juno is built on 25+ years of experience in human behavioural research. See how it works or get in touch.