I built an NPS tool and discovered the metric founders obsess over isn't the one that matters
Two years ago I was helping a small restaurant group figure out why one of their three locations kept underperforming at lunch. Same menu. Same prices. Same training. They'd tried surveys before, a generic SurveyMonkey link on a table tent pulling a 3% response rate, mostly complaints from people already annoyed enough to scan a QR on their way out.
We swapped in a branded NPS survey sent via SMS and email. Three weeks later they had 200+ responses and a score of +18. Not great. But the score wasn't the finding. Forty-seven comments mentioned the Tuesday lunch wait time. Not the wait itself. The gap between the quoted wait (12 minutes) and the actual one (often 30+).
They fixed one thing. The host stopped guessing and started quoting the real kitchen queue. Three months later NPS was +51 and lunch revenue at that location was up 22%.
That whole sequence eventually turned into NPSKit (npskit.com). But the more interesting thing for this audience isn't the product. It's what I learned about response rates while building it, because I think it applies to almost every founder running surveys, asking for feedback, or trying to listen to users.
The thing nobody warned me about: the tool barely matters. The branding does.
When a customer gets a survey link from a SaaS company they've never heard of, their brain categorizes it in roughly 400 milliseconds. Spam-folder behavior. Not because they're unfriendly. They'd probably give feedback happily if asked in a way that felt real. It's that nothing about the experience signals "this came from the place I just visited."
Across the businesses I ran this on, branded surveys pulled 3 to 4x higher response rates than generic third-party links. That's the difference between a dataset you c
Sharp story. The "47 comments named the wait time gap" finding is the real unlock buried in the middle.
Two things worth pushing on though. The 3-4x lift isn't really branding — it's identifiable context. "From the place I just visited" works because the customer remembers the interaction. SaaS surveys for a weekly-use tool hit the same response rate even unbranded. Variable that moves response rate is recency + memorability, not branding aesthetics.
Deeper insight your post almost surfaces but doesn't extract: the NPS score didn't matter. The 47 comments did. The restaurant fix came from open-ended responses, not the score. Your tool's real value prop is "branded delivery + comment analysis," not NPS specifically.
You could have found the wait-time problem with one open-ended question and no NPS at all.
(Post got cut off — curious where the rest landed.)
The most useful finding in the restaurant story isn't the score change, it's that 47 people independently mentioned the Tuesday wait time gap. That kind of clustering in open-text responses is usually the highest-signal data in any feedback system, and NPS scores often bury it because founders track the number and skim the comments. The 22% lunch revenue bump came from fixing a single operational behavior that 47 people mentioned, not from improving a score. Curious how NPSKit surfaces that pattern: does it do any keyword clustering on open-text responses, or is finding the 'Tuesday problem' still something you do manually by reading through comments?
This is a strong insight because the real product is not “NPS collection.” It is turning customer feedback into something businesses actually trust, recognize, and act on.
The branding point you made is exactly where I think NPSKit may have a ceiling. It explains the tool, but it also keeps the product tied tightly to one metric. Your own story shows the bigger value was not the NPS score. It was operational intelligence: finding the Tuesday lunch wait-time gap, fixing one behavior, and moving revenue.
That is much broader than an NPS kit.
If this keeps expanding into feedback, customer signals, branded surveys, comments, and business improvement actions, I’d pressure-test the name before more customers and content lock it into “NPS tool” territory.
Beryxa .com would fit that bigger direction better as a serious customer intelligence and decision platform. It would make the product feel less like a survey add-on and more like a business signal layer that helps operators know what to fix.