Audience measurement is the counting of how many people watched each TV programme, each advertisement break and each stream, and who those people were. The results are called ratings. Advertisers and TV networks use ratings to set the price of advertising. This page uses the United States as its example, where one company's ratings have long been the standard that advertising is bought on.
Nobody can watch every home. So the measurement company recruits a panel of tens of thousands of households. The panel is chosen so that, together, it looks like the country by age, income, region and other traits. Each panel home gets a meter attached to its TVs. The meter detects which channel or app is playing, often by picking up a code that the broadcaster places in the sound, too faint for people to hear. A people meter also asks each viewer to press a button on a remote, to say who is watching. Every night, the meters send the day's data to the company. Statistics then scale the panel up to the whole country. The method is explained in consumer panels and retail audits.
A panel alone is now too small. Viewing is split across hundreds of channels and streaming apps, and a small programme may be watched in too few panel homes to measure. So measurement companies add what they call big data: viewing records from millions of set-top boxes and smart TVs. The panel tells them who is watching. The big data tells them, in finer detail, what is being watched. Streaming services, connected TV (CTV) advertising and addressable TV advertising add their own records of which ads were played.
Take, for example, a Sunday evening game of American football shown on a US network. The next morning, the ratings estimate that 24 million people watched the average minute, and that 9 million of them were aged 18 to 49. Advertisers had bought their 30-second slots with a promise of a certain number of viewers aged 18 to 49. Prices are quoted per thousand viewers, called CPM, or cost per mille. If fewer people watched than the television network promised, it gives the advertiser extra slots free of charge. These are called make-goods.
Brands also join ratings to sales data from retail measurement, to see whether the homes that saw an advertisement bought more of the product. Other countries have their own ratings systems, run on similar panels.
How is it different from ad verification? Audience measurement estimates how many people, and which kinds of people, watched a programme or an ad break. Ad verification checks each single digital ad: whether it appeared on a screen, and whether a person rather than a bot saw it.