Read the sample size first
Every number we publish travels with the sample it was measured on. That is not a disclaimer — it is the part that tells you whether the number means anything.
A percentage on a marketing page is the easiest thing in the world to write and the hardest thing to check. Ours travel with their sample size for one reason: without it, a figure is not a measurement, it is a mood.
Two numbers, same shape, different weight
Consider two rows on our own product page. One reads 96.5% · n = 292 samples. The other reads 100.0% · n = 2 samples.
The second number is larger. It is also worth almost nothing, and we mark it as thin evidence so that nobody has to work that out for themselves. A hundred per cent of two is a coin landing heads twice.
Published as bare percentages, those rows would read as a strong result and a perfect one. Published with their samples, they read as what they are: one measurement we would defend, and one we would not.
Why not simply drop the weak rows
Because a selection is a form of fabrication. A table showing only the rows that flatter is a lie told entirely with true numbers, and it is the easiest kind to tell without ever writing a false sentence.
So the rule is blunt: a number never appears without its sample in the same block. Not in a footnote, not in a tooltip, not behind a link. If the two cannot be shown together, the number does not get published at all.
What it costs
Some of our figures look weaker this way, which is the intended outcome. A buyer comparing vendors can tell a measured claim from a confident one, and the buyers who cannot are not the ones this product does well with.
The alternative is a number that survives the sales conversation and dies in the first week of use. That trade has never once been worth making.