The Two-Question NPS: Why Most 'Add More Context' Advice Fights the Point of NPS

Net Promoter Score works because it's simple: one question, one number, one scale everyone uses the same way — "how likely are you to recommend this to a friend or colleague," zero to ten. That simplicity is precisely what makes it comparable across time and across companies. And yet a surprising amount of advice out there tells you to "improve" your NPS survey by bolting extra context onto the rated question itself — adding sub-clauses, qualifying it with "considering price and support," splitting it into multiple related ratings. Nearly all of that advice works against the one thing NPS is actually good at.

What breaks when you overload the rated question

The moment you add conditions to the NPS question, you've changed what the number means, and you've usually changed it in a way that makes it incomparable to your own past scores. "How likely are you to recommend us" and "how likely are you to recommend us, considering our recent pricing changes" are different questions with different baselines. Track the second one over six months and you can't cleanly tell whether a dip is about the product or about pricing — the question conflated them for you.

The people who reach for this kind of padding are usually trying to solve a real problem: a bare number doesn't tell you why someone gave it. That's a legitimate gap. It's just the wrong fix.

The fix is a second question, not a bigger first one

The cleaner way to capture the "why" is a single open-text follow-up question directly after the rated one — something as plain as "what's the main reason for your score?" This keeps the NPS question itself untouched and comparable over time, while still getting you the qualitative color that makes a score actionable instead of just a number on a dashboard.

This works because a rating question and a long-text question are structurally doing different jobs, and SparkyLoop's question types reflect that split rather than trying to merge them: NPS and rating questions produce a number you can average and trend; short and long text questions produce language you actually have to read. Keeping them as two separate questions in the builder — rather than trying to cram qualifying language into the rated one — is what lets you track the number cleanly while still hearing the reasoning behind it.

What you actually do with the follow-up answers

The value of the open-text follow-up isn't in reading every single response individually forever — it's in being able to when a number moves. If your NPS score drops five points in a quarter, that's the moment to actually read through the open-text answers from that period, especially from the detractors, and look for a pattern. Without the follow-up question, a score drop tells you that something changed; with it, you often get a decent read on what.

For surveys with enough open-text responses accumulated, SparkyLoop's AI Theme Summary (available on plans with AI Insights) turns a pile of open-text answers into a plain-English summary of the recurring themes — what people praised most, what came up as a complaint most often — once there are enough responses to summarize meaningfully. That's a fast first pass across fifty comments, not a replacement for reading the handful of outliers yourself, particularly the lowest scores, where the specific wording tends to matter more than the theme.

When a third question is actually worth it

None of this means an NPS survey has to stay at exactly two questions forever. If there's a second thing you genuinely intend to act on — say, a specific feature you're deciding whether to invest in — a third, separate question aimed squarely at that is reasonable. The line to watch is whether you're adding a question because you have a real, specific use for the answer, or because it feels incomplete to only ask two things. The second reason is exactly how a clean two-question NPS survey turns into the bloated, low-completion instrument it was trying to avoid being.

How to do this in SparkyLoop

Start from the NPS template, keep the rated question exactly as-is, and add one long-text question immediately after it asking for the main reason behind the score. Resist adding conditions to the NPS question itself. Build it in the survey editor — see the full set of question types on SparkyLoop's features page.