You run fourteen conversations. Twelve of them line up. Two do not. The natural instinct is to treat the twelve as the finding and the two as noise — wrong persona, bad fit, a distracted respondent. Sometimes that is true. Often it is the most expensive assumption in the whole set.
Why outliers carry weight
A majority pattern usually tells you what is normal inside the segment you recruited. It does not tell you whether that segment was the right one. The two conversations that broke the pattern are the only place in your data where a different reality showed up — a different constraint, a different buyer, a different way of solving the problem today.
- They may be a different segment you have not defined yet.
- They may be further along, and describing the problem you will hit in a year.
- They may have already solved it with a workaround — which is real competition.
- They may be the only ones with budget and authority to actually buy.
How to work an outlier
Do not average it away. Go back to the transcript and read it on its own terms. What was different about their situation, not their opinion? What did they do, not what did they say they would do? Then go find two or three more people who look like them and run the same conversation again. Either the pattern reappears — in which case you have discovered a segment — or it does not, and now you can set it down with a reason.
Consensus tells you where you already are. The outlier tells you what you have not looked at.
The failure mode this prevents
Teams that only count agreement tend to build a product that is well matched to the people who were easiest to reach. The outlier interview is the cheapest early warning you will get that the easiest people to reach are not the people who will pay.