"Sweeps" are an institutional feature of American television
broadcasting by which TV viewership is measured four times each year ("sweeps periods"), in November, February, May, and July. This viewership data is then used to set the price of advertising airtime, giving broadcasters a strong incentive to inflate their audience during sweeps periods.
There are a variety of ways you might feasibly money from broadcasting.
If you sell broadcast receivers like radios or television sets, you might conduct your own broadcasts as a marketing tool to induce consumers to purchase the equipment necessary to tune in. This was the basis on which RCA initially operated the NBC radio and television networks.
You might acquire a charter from the government and collect taxes or user fees from television owners. This is the way the BBC and many other state-affiliated broadcasters are funded.
You might solicit donations from viewers, most commonly through televised "telethon" appeals occurring in place of regularly scheduled programming. Americans might be familiar with this tactic from broadcasters like the Christian Broadcasting Network or various local public television stations.
If you are an American broadcaster, however, you most likely make most or all of your money by selling airtime to advertisers who want to say something to your audience, most commonly in support of a commercial product or service. A typical American television broadcast, as of time of writing, contains sixteen minutes of advertising in each hour of programming.
So, how much can you sell this airtime for? Well, advertisers will obviously prefer a bigger audience, or an audience more favorably disposed to their message. And the bigger and better an audience you can give them, the more they'd be willing to pay for your airtime. But here a problem arises: how exactly do you know who, or how big, your audience is? Broadcast media, unidirectional by nature, provided no automatic feedback, and it would be impossible to measure viewership directly.
So, sampling presented itself as an alternative, and many early attempts at solving the problem involved phone surveys, in which (ideally) random households were called and inquired about their viewing or listening patterns. This, of course, raised serious issues of selection bias, and was generally unsatisfactory. Surveys directing respondents to list all recent viewership often returned inaccurate records. Surveys that limited their inquiry to the television program the respondent was currently watching were a bit more accurate, but even in an age of channel listings in the mid-one-digit range, to collect data of any depth with this method would be exceedingly manpower-intensive.
By the 1950s, marketing research firm A.C. Nielsen Co. was already using a mechanical recorder, the Audimeter, to track television viewership. However, the uneconomical awkwardness of the device (users had to remove and mail a film cartridge to Nielsen each week) limited Audimeter use to only a few hundred households nationwide, a sample large enough to reasonably measure the viewership of nationally broadcast network offerings but too small to assess viewership in individual broadcast markets. Furthermore, the device could not record any more detailed data about the audience, such as the number of viewers watching a given program, or their demographic makeup.
In 1953, however, Nielsen tried supplementing the Audimeter with written logs of programs watched, which allowed for more extensive surveying, on the order of hundreds of thousands or even millions of households. The test was a success, the data proven reliable. Due to the costs associated with processing these diaries, however, it was decided to employ this technique on a broad scale only periodically.
1954 thus saw the first sweeps, which set the pattern for all subsequent instances. Nielsen mailed diaries to households throughout the country, who recorded a week's worth of television viewing. Nielsen then collected the diaries, the whole process staggered by region, beginning with the American Northeast and then subsequently the South, Midwest, and finally West. This characteristic "sweeping" collection pattern gave sweeps their name.
Sweeps have remained fundamentally similar since then, though the demographic data collected has grown a bit more complex. Currently, around 1.6 million households maintain sweeps diaries each year. These "Nielsen families" are paid a token amount; most households presumably participate so as to influence programming in accordance with their preferences. If you're an American, you probably know someone who kept such a diary, but Nielsen is quite adamant in insisting that Nielsen families keep their identity a secret.
Given that sweeps measurements are only taken quarterly, during known dates, it is possible for broadcasters to "game" the system and temporarily boost ratings. Sweeps are traditionally full of attention-whoring "sweeps programming". To headline the event, networks offer up miniseries, original movies, or TV debuts of big-screen blockbusters. Regular series get in on the game too, featuring big-name guest stars, major plot twists, or shocking and thus "must-see" episodes (same-sex relationships, for example, have been heavily employed in this capacity recently, starting with Ellen's pathbreaking "Puppy Episode", during the the May 1997 sweeps). Sweeps periods frequently see a shakeup in the network lineup as underperforming shows are canned, or at least put on "hiatus" to make way for midseason replacement debuts, reruns of more popular series, or the more favorable placement of remaining shows.
Nonfiction programs are not exempt from all this hoopla, as local and network news programs will frequently run interviews with major figures, or stage dramatic stunts of questionable news value. News programs will also frequently be drafted into service promoting the channel's other offerings, featuring interviews with television show cast members or guest stars, reports on the topics on which fictional offerings are based, previews of big-name events, or simply stories about the current sweeps themselves.
This is all admittedly a little disingenuous, but keep in mind that sweeps ratings are not completely disconnected from year-round performance. Viewers won't know about your must see spectacular if they don't see the network promos airing every other commercial break. Viewers won't care if your series' lead character gets pregnant/married/shot if they don't know or care who the character is. Channel surfers won't get sucked in by your screening of the latest Jerry Bruckheimer movie if they don't think to check in on your station in the first place. Sweeps build on viewer loyalty built up during the non-sweeps season. A solid sweeps showing allows you to use sensationalism and canny promotion to build on a solid foundation, not in place of such fundamentals.
Sweeps and Advertising
It is a common misconception that all television advertising rates are set in accordance with sweeps week ratings. In fact, only local advertising rates are affected. For network television, usually only 4 of the aforementioned 16 minutes of advertising in each hour of television are local ads, purchased from individual stations. The remaining 12 minutes are national advertising, purchased from the national network in accordance with weekly ratings collected from around 5,000 households by the "People Meter", the Audimeter's less unwieldy successor. Network ad rates can thus be boosted only through consistent quality programming, but the desire to help out network-owned stations and keep affiliates happy keeps networks focused on maximizing sweeps ratings.
In theory, it might be possible to expand year-round People Meter use to such a point where it could supplant sweeps diaries in measuring local ratings, giving local advertisers a more accurate picture of stations' viewerships. However, such a plan could be expected to draw resistance from networks and stations (fearing decreased revenues), ad agencies (whose commission-based contracts give them a vested interest in high costs), and major national advertisers (who appreciate the way in which the system weighs against small, upstart competitors). Compared with these groups, local advertisers (usually small, dispersed, and unorganized) have very little pull in the media world.