The file arrived with no country field. This is the exact moment most reports quietly invent one, and it is the moment BeanBreaker decided, on principle, not to. What follows is how you place two-thirds of a world without ever once guessing – and, just as importantly, why the guessing is precisely the part you have to refuse, even when refusing makes your headline number smaller.
A dataset without a location column is not a dead end. It is a room full of fingerprints. People leave their country all over their profiles without ever naming it directly – in the shape of a phone number, the tail of a web address, a flag they tapped once and forgot, the very script their sentences are written in. Nobody types “I am in this country” into a coffee bio. But almost everybody leaves the evidence anyway, scattered and unlabelled, waiting for someone patient enough to read it in the right order. The work behind How The World Talks About Coffee, August 2026 was to read those traces in a fixed sequence of confidence, because a good method is, in the end, mostly a good sequence. You never weigh a hunch against a fact. You reach for the strongest available signal first, and you only fall back to a weaker one when the stronger one stays silent.
So the priority ran like a well-pulled shot, each stage feeding cleanly into the next. First, international dialling codes – the hardest, least ambiguous evidence on offer – with North American area codes resolved individually, one by one, so that a Canadian or a Caribbean number was never lazily swept into the United States pile for convenience. That single act of discipline did the heavy lifting: 65.4% of all placements, 1,038 accounts, settled on the strength of the phone number alone. Where no number spoke, a place name written into the biography itself carried the next tranche – 14.9%, or 236 accounts. Below that came website domains for 8.8% (140 accounts), then email domains for 5.6% (89), then a flag emoji for 4.9% (77), and finally, for the last stubborn handful that offered nothing else at all, the alphabet itself – the script a person chose to type in – for 0.4%, a mere 7 accounts. Each tier caught what the tier above it had missed. None was ever permitted to overrule a stronger signal sitting above it in the order.
The result of all that patience is 1,587 of 2,164 accounts placed across 86 countries – 73.3% of the entire file, reconstructed from evidence that was always sitting there in plain sight but had never once been labelled. Nearly three-quarters of a location-less dataset, given a home, without inventing a single one. Every one of those 86 countries was earned off a real trace a real person had left behind, not conjured to fill a gap in a map that looked too empty to publish.
And here is the part that separates genuine research from confident decoration, the part worth reading twice. The remaining 577 accounts are reported, plainly, as Unknown. Not smeared proportionally across the map to make the totals look rounder and more complete. Not quietly assigned to whichever country would have flattered the headline geography. Not filled in with the statistical equivalent of a shrug. Left blank, deliberately, on the record – because an honest gap is worth more than a confident fabrication every single time, and because the reader deserves to know the precise coordinates of where the knowing stops and the guessing would have begun.
That, in the end, is the whole ethic of the thing, and it is exactly the ethic you would want in anyone you ever hired to find something out on your behalf. Extract every real signal the evidence actually contains. Order those signals honestly by how far each one can be trusted. And when the signals finally run dry, as they always eventually do, say so out loud rather than papering over the silence. A study is never judged by how much it claims to know. It is judged, every time, by whether you can trust the line it draws between what it genuinely found and what it left honestly, unglamorously, and unapologetically blank.
