There is a sentence that gets said in a lot of rooms, and it always sounds like rigor. We can't act on that yet. We don't have statistical significance.
Sometimes that is true. Most of the time, in my experience, it is a stalling technique wearing a lab coat.
The tell is the number
Notice where the threshold always lands. You need a hundred customers before we can map this. You need six months of data. You need the campaign to be forty percent complete.
It is always more than you currently have. Every time.
Which is the tell, because if the number were real, someone would eventually have it, and then the work would have to start. A threshold that stays permanently just out of reach is not a measurement standard. It is a deferral mechanism, and it is doing exactly the job it was built to do.
If you already had the number, they would have to actually help you.
I am not accusing anyone of dishonesty. The people who say it usually believe it, because it is what they were taught to say when they are not yet sure what to do. The honest version of the sentence is I don't know what to recommend yet, and that sentence is fine. Nobody gets fired for it. But it is harder to say than the one about significance, so the one about significance is the one that gets said.
What actually validates something
Convergence, not volume.
If twenty customers tell you the same thing, unprompted, in their own words, that is validated at twenty. Not provisionally. Validated. You may learn something new at fifty that changes your mind, and you should stay open to that, but the signal at twenty was real and acting on it was correct.
The reverse is also true and less comfortable. If two hundred customers tell you eleven different things, you do not have significance either. You have volume and no pattern, and no amount of additional volume will manufacture one. Volume is not the variable that determines whether you know something. Agreement is.
So the question is never do I have enough customers. It is are they saying the same thing. That question you can answer this afternoon, at whatever count you are at.
What waiting actually costs
Here is what makes this expensive rather than merely annoying. The delay is not neutral. While you wait for a threshold that recedes as you approach it, every one of those customers keeps buying, not buying, staying, or leaving on the current model, and every quarter of that is a quarter you did not spend acting on what you already knew.
This shows up everywhere, but it shows up hardest in expansion, because expansion is the discipline most easily postponed. Nothing goes red when you skip it. Nobody misses a target that was never set. So we need more data first is the easiest sentence in the world to say about it, and the most expensive one to believe.
You do not need a hundred customers to write down what your customers could buy from you. You do not need statistical significance to notice that the last four who left had all achieved something first. You need to look at what you have and be willing to be wrong in public later.
Take the data you have to the strongest conclusion it honestly supports, act on it, and let the next fifty customers correct you. That is not recklessness. That is the only version of this that ever produces anything, and the money it produces starts arriving during the quarters you would otherwise have spent waiting.
