Cost Per Mille: The Common Currency of the Ad Markets

A Latin word for a universal price

Cost per mille takes its name from a Latin word, mille, meaning a thousand, and it names something almost mundane: the price an advertiser pays to reach one thousand views, or impressions, of an advertisement. The term is old enough that its Latin root sits oddly beside the industries that now use it. A CPM is quoted for a radio spot, a page in a magazine, a poster on a highway, and a banner ad on a website, in the same units, the same way a shopper compares two products by price per kilogram rather than by the size of the packaging.

The arithmetic behind it is plain. Divide the total cost of an ad placement by the number of impressions it delivers, then express the result per thousand impressions rather than per single view, since a single view is too small a unit to price sensibly. What the arithmetic does, though, is not plain at all. It lets a media buyer set a price per thousand exposures for a full page in a national newspaper and set the same kind of price for thirty seconds of drivetime radio, then ask which number is smaller.

Before that comparison was possible, print, radio, television, and outdoor advertising were, in effect, separate economies. Each sold a different unit of value, column inches, ratings points, poster faces, and each was bought by a different specialist who understood that medium's own conventions and none of the others. Cost per mille did not change what any of those media were. It changed what could be compared. Once every medium's audience could be restated as a price per thousand, a dollar earmarked for advertising stopped belonging to one medium and became free to move to whichever medium's thousand cost the least.

The comparison did more than help a single buyer choose a channel. It meant that print, radio, television, and out of home were, for the first time, competing suppliers inside the same market rather than separate industries that happened to share the word advertising. That reframing, of what had been distinct trades into one continuous market for attention, is the quiet argument this history makes, and it played out again, on a compressed timeline, when the web adopted the same unit a century's worth of publishers and broadcasters had already agreed on.

Before the common denominator

Cost per thousand pricing did not arrive with the internet. It originated in print and broadcast rate card practice, the standard method publishers and stations had used for decades to sell advertising space and airtime before a browser existed. A newspaper's or a broadcaster's rate card already listed a price against a thousand copies sold or a thousand viewers reached, so that when digital advertising adopted the same convention in the 1990s, it was reusing an existing pricing logic rather than inventing a new one. That reuse is why a magazine's CPM in 1993 and a website's CPM in 1998 could be set side by side and understood immediately by anyone who bought both.

That convention rested on one condition: that the thousand being sold was real. Early circulation claims were not always trustworthy. Publishers had every incentive to inflate the number of readers they were charging advertisers to reach, and for years there was no independent check on the figure. The Audit Bureau of Circulations was built in 1914 specifically to solve that problem, verifying circulation so that the denominator inside a rate card, and later inside a CPM, could be trusted rather than taken on a publisher's word.

The audited circulation model cut both ways, which is worth holding onto because the same tension follows the metric into every medium that later adopts it. Independent verification let advertisers trust a number they had not personally checked, and it let smaller or newer publications compete on a comparable, honest footing against the giants of their day. But it also concentrated authority in whichever body ran the audit and in whichever publishers could afford, or agree, to be measured by it. A denominator that can be trusted is also a denominator someone controls.

The banner that is misremembered

The web's advertising history usually opens with a single date: October 27, 1994, when Wired's new online offshoot, HotWired, ran a banner for AT&T's "You Will" campaign, widely credited as the first banner ad on the internet. The deal was struck in San Francisco, inside the same stretch of Northern California that was already becoming the center of the American technology industry, and it was AT&T's Silicon Valley agency relationships, not any international standards process, that shaped how the sale was made.

The banner performed in a way no digital ad has matched since: 44 percent of the people who saw it clicked on it, a click-through rate roughly one hundred times higher than the rate a typical banner ad would draw a decade later, once cost-per-thousand-priced inventory had scaled into an ordinary, oversupplied commodity. Attention that novel and that scarce could command almost any price, which is itself evidence for the argument that follows: once a market for a thousand exposures becomes large and comparable, the price of any single exposure inside it falls.

What the standard telling usually leaves out is how AT&T actually paid for that first banner. The deal was not priced per thousand impressions or per click. It was a flat fee, thirty thousand dollars, for a fixed placement period of three months, the same magazine-style logic publishers had used to sell a page for a set span of issues. The advertisement remembered as the origin of online, per-impression pricing was, in fact, sold the old way, before cost per mille and its offspring, cost per click, displaced that flat-fee logic on the web within a few years.

The correction matters for what it shows rather than for the trivia itself. Cost per mille did not arrive on the web with a single founding transaction. It migrated, the way a currency migrates into a new market once enough buyers on both sides already understand it, and the migration happened quickly because print and broadcast had already spent decades teaching advertisers, agencies, and now internet publishers to think in that unit.

One price, one market

Once cost per mille was applied consistently to web advertising, in the banner exchanges and early ad networks that followed HotWired through the second half of the 1990s, it did for the internet what it had already done for print against broadcast: it let a single media budget shop across formats that had nothing else in common. A media buyer no longer had to decide in advance whether the money would go to television or newsprint or a website. The buyer could set a target price per thousand exposures and let the number decide, moving the dollar to whichever channel's thousand cost the least that quarter.

That is the quiet, structural change cost per mille made possible: newsprint, airwaves, billboards, and banners, four industries that had spent most of a century selling entirely different products in entirely different units, became four suppliers competing inside one continuous, arbitrageable market. A publisher did not need to know anything about broadcast ratings to know it was losing budget to a radio station whose CPM undercut its own, because the number that mattered was expressed in the same currency on both sides.

A unit that rewards reach at any cost, however, is not the same as a unit that rewards results, and that gap is where cost per mille began losing ground almost as soon as it had unified the market it created. Publishers favor CPM because it is paid regardless of what happens after the impression is served. Advertisers, once they could measure what happened after a click or a purchase, increasingly preferred to pay only for that measurable outcome, through cost-per-click and cost-per-acquisition pricing instead. The shift toward performance pricing is, in effect, bargaining power moving from the seller of attention to the buyer of it, inside the very market cost per mille had made comparable in the first place.

Both readings of cost per mille are true at once, and neither cancels the other. It freed capital that had been locked inside separate media silos, letting a dollar chase the cheapest attention wherever that attention happened to be sold, a real gain for advertisers and for smaller publishers who could finally compete on price rather than reputation alone. It also concentrated power in whoever could guarantee volume at scale and set the rules for how a thousand exposures would be counted, which tends to favor the largest networks and platforms over the smallest ones. A common currency does not distribute the gains from trade evenly just because it makes the trade possible.

Who wrote the meter

A price convention needs an authority behind it, someone whose count of a thousand impressions the rest of the market agrees to trust, the same role the Audit Bureau of Circulations played for print decades earlier. For the web, that role fell largely to the Interactive Advertising Bureau, an American trade body formed by the country's digital publishers and ad networks, which formalized the standards, including CPM measurement conventions, that the online advertising industry would use going forward.

The consequence outlived the era of banner ads. A body that sets the counting rules for an entire market's pricing unit also has a stake in defending the conditions that keep that unit valuable, and the IAB has, in fact, lobbied against strict online privacy legislation, arguing that tighter rules on data collection would damage the digital advertising economy its members depend on. The connection to cost per mille is direct: better-targeted advertising commands a higher CPM, so restrictions on the data that makes targeting possible threaten the very price the trade body's own convention exists to set. A pricing unit invented to compare attention across media had, within a generation, become an argument used inside privacy policy debates well beyond advertising alone.

The meter's blind spot, and what may replace it

Cost per mille is sometimes presented as advertising's honest number, the figure that finally let every medium be judged on the same, transparent scale. That framing understates a limitation the metric has carried since it left print and broadcast. For media without a countable exposure, an outdoor billboard, or certain radio formats, a CPM cannot be directly measured; it has to be estimated from research on likely audience size instead. The common currency of the ad markets, in other words, rests in part on modeled figures rather than audited ones, a structural echo of the same circulation-fraud problem the Audit Bureau of Circulations was built to solve in 1914. Verification, it turns out, is a job that has to be redone for every new medium, not solved once for all of them.

The newest medium raising that question is the one this publication measures for a living. When a buyer asks an AI system which business to trust and the system names one, that placement is not sold, or counted, in thousands of impressions the way a banner ad or a radio spot is. There is, at least so far, no settled per-mille price for being named inside a generated answer, no rate card, and no trade body that has formalized how such a mention would be counted the way the IAB formalized the online impression. Whether a comparable common denominator eventually forms around answer-engine placement, and whether it forms under an American convention the way cost per mille did, is an open question rather than a settled one, worth watching rather than predicting.

What cost per mille demonstrates, regardless of how that next chapter turns out, is that a market does not become a single market by accident. It becomes one when a unit is invented that lets otherwise incomparable products be priced against each other, and the group that gets to define how that unit is counted quietly holds a kind of authority over the resulting market that outlasts any single medium, deal, or campaign inside it.