[{"data":1,"prerenderedAt":140},["ShallowReactive",2],{"blog-\u002Fblog\u002Ffrom-sofas-to-viewing-figures":3},{"id":4,"title":5,"author":6,"body":7,"coverImage":123,"description":124,"draft":125,"extension":126,"meta":127,"navigation":128,"path":129,"publishedAt":130,"readingTime":131,"seo":132,"stem":133,"tags":134,"__hash__":139},"blog\u002Fblog\u002Ffrom-sofas-to-viewing-figures.md","From a few thousand sofas to the nation's viewing figures","Evan Ritter",{"type":8,"value":9,"toc":112},"minimark",[10,14,17,20,25,28,31,34,38,41,44,47,51,54,57,60,63,67,70,73,77,80,83,86,89,93,96,99,102,106,109],[11,12,13],"p",{},"Whenever someone finds out what I do for a living, the same question tends to surface within about thirty seconds. It happened again last week, and it was actually my wife who pointed out that I have a sort of stock answer I trot out every time — and that maybe I should just write it down. So here it is.",[11,15,16],{},"The question is always some version of: \"When they say six million people watched something, how do they actually know? Is there a box in my telly counting?\"",[11,18,19],{},"It's a fair question, and the honest answer is more interesting than people expect. Nobody is counting all of us. The whole thing rests on a sample — a few thousand homes — and a surprising amount of statistical machinery to turn that handful of living rooms into a number that stands in for the entire country. Here's the journey, from the sofa to the figure you read in the morning.",[21,22,24],"h2",{"id":23},"it-starts-with-a-panel-not-a-headcount","It starts with a panel, not a headcount",[11,26,27],{},"There is no master meter wired into every television in Britain. Instead, viewing is measured the way most big population questions are answered: with a carefully chosen sample.",[11,29,30],{},"A panel of several thousand homes — recruited and maintained to mirror the UK population — does the watching on everyone's behalf. The homes aren't picked at random off the street. They're balanced to match the country across the things that shape viewing: age, gender, social grade, region, household size, whether there are children, what platforms the home has, and so on. The goal is that the panel, in miniature, looks like the nation.",[11,32,33],{},"Keeping that mirror accurate is a job in itself. The make-up of the population is checked against a large, separate survey of households, and the panel is continually topped up and rebalanced so it doesn't drift away from reality as homes join, leave, or change.",[21,35,37],{"id":36},"measuring-who-is-watching-not-just-what","Measuring who is watching, not just what",[11,39,40],{},"Here's the part most people haven't thought about. It isn't enough to know a television was switched on and tuned to a particular channel. To say \"six million adults\" or \"the audience skewed young,\" you need to know who was actually in the room.",[11,42,43],{},"So in a panel home, it isn't only the set that's metered — the people are accounted for too. Each member of the household has a way of registering that they're present and watching, and guests can be logged as well. That's the secret behind every demographic breakdown you've ever seen: the system isn't inferring who watched from the channel, it's recording it, person by person, minute by minute.",[11,45,46],{},"That granularity is why a broadcaster can be told not just how many people watched, but how many women aged 16–34, or how many ABC1 adults, or how the audience built and fell away across the half-hour.",[21,48,50],{"id":49},"the-clever-bit-turning-a-sample-into-a-nation","The clever bit: turning a sample into a nation",[11,52,53],{},"This is the step that feels like sleight of hand until you see it, and it's the real answer to \"how do they know?\"",[11,55,56],{},"Every home and every person on the panel carries a weight — a multiplier that says, in effect, \"this household stands in for roughly this many households like it out there.\" A home that represents an under-sampled slice of the country carries a heavier weight; an over-represented one carries less. Add up the weighted viewers and the few thousand panel homes scale up to the tens of millions of people they're meant to represent.",[11,58,59],{},"The totals they scale up to aren't guessed, either. The size of each population group — what the industry calls the universe — is published and updated regularly, so everyone is multiplying up to the same agreed picture of the country. That's what converts \"412 of our panellists watched\" into \"an audience of X million,\" and it's why two people working from the same data arrive at the same figure.",[11,61,62],{},"It's worth pausing on how much rests on this. The reason a sample of a few thousand homes can speak for a whole country isn't magic — it's the same logic that lets a political poll of a couple of thousand people estimate a national vote. Get the sample representative and the weighting honest, and a small, well-built panel is astonishingly accurate.",[21,64,66],{"id":65},"and-then-theres-everything-that-isnt-live","And then there's everything that isn't live",[11,68,69],{},"Years ago this would nearly be the end of the story: measure the live broadcast, weight it up, done. Not any more. We don't watch in neat live blocks. We record things, we start a box-set an hour late, we finish the episode on a catch-up app three days afterwards on a tablet.",[11,71,72],{},"So the measurement has had to grow to follow the viewing wherever it goes. Watching on the streaming players is now folded in too, with the panel calibrated against data coming from the players themselves — which means the figure for a programme increasingly captures the people who watched it days later on an app, not just the ones who sat down for the live transmission.",[21,74,76],{"id":75},"the-overnight-rush","The overnight rush",[11,78,79],{},"This is the bit I find genuinely satisfying, because it's as much logistics as statistics.",[11,81,82],{},"A viewing \"day\" doesn't run midnight to midnight — it runs from the early hours through to the early hours of the next day, so that someone still up watching at one in the morning is counted against the right day. In the small hours, while the panel sleeps, their meters are quietly polled and the night's data is collected in. It's then processed at speed, and the first set of figures — the overnights — is ready by the middle of the next morning.",[11,84,85],{},"Those overnights are the numbers the industry lives and dies by at breakfast: the provisional, live picture of who watched what the day before. But because they land so fast, they can only include viewing that had actually happened by then — essentially the live audience plus same-day playback.",[11,87,88],{},"So the figures get a second life. About a week later a consolidated set arrives, this time including up to seven days of recorded and on-demand viewing layered back onto each programme. A programme that looked modest overnight can grow noticeably once a week of catch-up is added. There's even a later pass that stretches the window out to 28 days. A viewing figure, in other words, isn't a single fixed thing — it firms up over days and weeks as the slower ways we watch are counted in.",[21,90,92],{"id":91},"where-the-bureau-comes-in","Where the bureau comes in",[11,94,95],{},"There's one last link in the chain, and it's the one I spend my days in.",[11,97,98],{},"What comes out of the overnight processing isn't a tidy report. It's a stack of dense data files — dozens of different types, each covering a slice of the picture: audiences minute by minute, programme details, advertising spots, the commercial breaks, the on-demand layer, and more. They're not designed to be read; they're designed to be ingested. Open one in a text editor and you'll find rows of fixed-width numbers with not a word of explanation.",[11,100,101],{},"That's the gap a bureau fills. We take those raw daily files, load them into a database, stitch them together, and turn them into the things people actually use: the overnight ratings tables, the programme-by-programme audiences, the channel shares, the advertising spot ratings, the reach-and-frequency analysis a media planner needs. The measurement industry produces the data; the bureau turns it into answers.",[21,103,105],{"id":104},"so-when-they-say-six-million-watched","So, when they say six million watched…",[11,107,108],{},"…what they really mean is this: a representative sample of homes, where the people as well as the sets are accounted for, did the watching; their viewing was weighted up to an agreed picture of the whole population; the night's data was collected and processed before most of us had finished breakfast; the figure was then revised over the following days as recorded and on-demand viewing was added; and somewhere in there a bureau turned a wall of raw numbers into the sentence you read.",[11,110,111],{},"It's a small marvel, really — a chain of sampling, measurement, statistics and logistics that runs every single night, mostly invisibly, so that the morning can have its number. I rather like that the next time someone asks me, I can just send them this.",{"title":113,"searchDepth":114,"depth":114,"links":115},"",2,[116,117,118,119,120,121,122],{"id":23,"depth":114,"text":24},{"id":36,"depth":114,"text":37},{"id":49,"depth":114,"text":50},{"id":65,"depth":114,"text":66},{"id":75,"depth":114,"text":76},{"id":91,"depth":114,"text":92},{"id":104,"depth":114,"text":105},"\u002Fblog\u002Ffrom-sofas-to-viewing-figures\u002Fcover.png","How a panel of a few thousand homes — plus a lot of statistical machinery and an overnight scramble — becomes the morning's TV ratings, and where a data bureau fits in.",false,"md",{},true,"\u002Fblog\u002Ffrom-sofas-to-viewing-figures","2026-06-13",8,{"title":5,"description":124},"blog\u002Ffrom-sofas-to-viewing-figures",[135,136,137,138],"broadcasting","audience-measurement","statistics","data-pipelines","lR_zEjtcpGRYkkCyzWGQjv13xTyNyidtidSQdtGxiTI",1781363261390]