While a great deal of what I write here is underinformed speculation, this piece is unusually speculative and underinformed. It's possible that I'm flat out wrong about the idea I'm developing here. Im putting it forward with the hopes that folks will react with examples and data that help prove or disprove this theory. Being told that I'm unambigiously wrong with good data demonstrating my error would be very helpful. Simply being told I'm wrong - less helpful.

A friend works for the local newspaper, the Berkshire Eagle, editing a weekly arts and calendar section. I visited her at her office the other day and surprised - staggered, actually - at the size of the office. While the paper relies on dozens of unpaid correspondents and pay-per-piece freelance contributors, it also has eight editors, several reporters, professional compositors/paginators, and departments for administration, circulation and ad sales. My friend guessed that, with everyone in the office, day and night shift, there were at least sixty full-time employees responsible for producing, distributing and monetizing the paper.

I don't mean to suggest that this is excessive, or that the newspaper's (legendarily tightfisted) management is running anything other than a lean, efficient operation. While the office is responsible primarily for the 26,000 circulation daily newspaper (not bad for a paper that serves a county of 135,000 people), it also publishes a 6,000 circulation paper focused on the northern Berkshires and southern Vermont. And the mighty four-color presses in the basement produce an additional Vermont paper, helping defray the costs of production. (Given the parent company's financial difficulties, and the downgrade of their bonds deep, deep into junk status, any efficiencies are surely appreciated.)

What I was amazed by was how much advertising and subscription revenue the paper must generate to support dozens of full-time jobs. I haven't been able to find revenue numbers specifically for the paper, but a back of the hand calculation suggests that it must be generating significantly more than $2 million a year in subscription and ad sales to support salaries and production costs. It's quite possible that subscription is what keeps the paper alive - at a retail price of $176.88 for daily home delivery, subscriptions could generate a couple million a year. (I assume not everyone is paying retail price, which means that some significant part of the paper's revenue picture comes from advertising.) I find the advertising side of the business harder to comprehend, however.

I help run a fairly popular online citizen media site, which is visited by roughly 300,000 visitors a month. We're doing a great deal of strategic planning at the moment, looking for ways to broaden our revenue base from foundation support and corporate parterships to include online advertising revenue. As we run the models, the numbers aren't especially pretty. Suffice it to say, very few of our models project us making a million dollars a year in online ad sales. Fortunately, we've got less than half a dozen full-time employees as well as an army of volunteers and contractors, so we don't need nearly as much revenue to support our work as our local newspaper does.

The difference in ad pricing between online and print advertising is something I'm finding mind-boggling. Advertising inserts in the Berkshire Eagle are priced at a base rate of $45 per thousand customers for two print pages, targeted by zipcode. Those prices don't include production costs - the advertiser is responsible producing the inserts and delivering them to the production facility. The cost covers the insertion of the ad into the appropriate papers and their delivery.

This model of advertising pricing - called CPM (cost per thousand) - is what early internet companies tried to use to monetize content. In the early days of Tripod, we called our friends who worked for glossy magazines, found out what they were charging for ads in CPM terms - usually $20-$60 - and asked our advertisers to pay at least as much. Our logic? Since you can't click on a magazine ad, they simply build brand, while ours build brand and can lead to sales.

We were able to earn $45 CPMs for a little while, until advertisers started to question our logic. In some cases, they saw other sites offering much cheaper rates; in other cases, they questioned whether the ads really were leading directly to sales, as they could track who'd clicked on the ads. CPM rates fell across the industry. Some sites continue to make pretty good money on a CPM basis, especially sites with highly targetted content likely to appeal to a certain type of customers. Federated Media, an advertising network that focuses on technology content, offers CPM ads from $3 to $26. Most content providers don't see rates that high. According to Technorati's 2008 state of the blogosphere, US bloggers who accept advertising saw a mean CPM of $4.20, but a median of $1.20 - that implies a small number of bloggers earning high CPMs for highly targetted content, and a lot of folks selling impressions at $1 per thousand.

The real problem for web publishers like Tripod wasn't the fall in CPM rates, but the switch to "performance-based advertising". As advertisers started to wonder whether ads were really leading towards sales, they started to pay not on an impression basis, but on a click basis. This makes much better sense for an advertiser - a click is a sales lead, and a marketer can decide she wants 500 new leads a month and will pay for 500 clicks. The problem with pay per click from the content provider's point of view is that clicks depend a great deal on the creativity of the ad. A truly great ad might get clicked 2-3% of the time, while a lame one could get less than 1% clickthrough, eating valuable ad inventory.

Google and Overture both figured out how to navigate the pay-per-performance world, setting up ad auctions where advertisers could bid to place their ads on search results pages, paying per click. Ads that were rarely clicked were placed lower on the results page, resulting in fewer clicks and incenting the advertisers to improve their copy. While highly targeted ads (an ad for roofing services in Pittsfield, MA) might be worth several dollars a click, most ads sell for a dollar or less a click, often much less. An ad that sells for a buck a click and gets 1% clickthrough is functionally a $10 CPM ad, which suggests that lots of ad inventory (the nickel-a-click stuff) selling at sub-$1 CPM.

Pay per performance seems to be the way of the future. JP Morgan analyst Imran Khan sees performance ad revenue growing more quickly than impression based and representing roughly two-thirds of the online market. CPM-based advertising is apparently less appealing to advertisers and may fall in price.

That's why $45 CPM seemed pretty high to me. Not that there isn't ad inventory that might be worth the price. The New York Times has recently started selling ads on its front page for $75,000 on weekdays, a $75 CPM, given the paper's print circulation of slightly over a million. The front page of the national paper of record is a pretty good way to gain some attention for your product or service. But that's not what we're talking about - we're talking about the sort of inserts supermarkets put into your newspaper to promote a 3 for 1 sale on nectarines. Those tend to catch my attention only when I'm looking for kindling. It's hard to believe that they are, reader for reader, worth 60% of what a front page ad on the New York Times is worth. Or that they're worth paying $4.50 a click, which is what would yield a $45 CPM in a pay per click environment, assuming 1% clickthrough.

Perhaps these prices aren't entirely rational.

Scott Karp wrote a lovely piece for Publishing 2.0 about 18 months ago called "Newspaper Online vs. Print Ad Revenue: The 10% Problem". The title refers to Karp's astute observation that "print circulation is about 10% of total audience reach, while online advertising revenue is 10% of total ad revenue [~] the economics are nearly the perfect inverse of what they should be." In other words, online readers of the Times were worth roughly 1/100th of what offline readers were in advertising terms. (They're worth even more inasmuch as most are paying subscription fees.) There's all sorts of ways to make these numbers make more sense - circulation bureas estimate that issues of the Times are read by multiple people, which means each of those print ads viewed multiple times. But the fact remains - online ads sell for much, much less than offline ones.

The comment thread associated with the piece is helpful, and includes a number of readers trying to help explain the disparity. Online advertising isn't especially mature yet, one suggests, and therefore significantly cheaper. The ability to target specific geographic markets makes newspapers more useful for non-virtual retailers. (Hmm. Given that the Times has been losing local audience and becoming a national paper over the past decade, this one is harder to swallow.)

A comment from Joshua Jeffryes rings true to me:

When I worked in Advertising the ineffectiveness of advertising was hardly a secret. But customers couldn't measure the effectiveness of ads. So they paid and continue to pay ridiculous prices for them.

Online ads, on the other hand, are measurable. They work just as well, if not better, than print, television, etc., the difference is that for the first time ad customers know exactly how ineffective they are.

Basically, there are two ways to explain the disparity in online and offline ad cost. One is to argue that paper ads are, for some combination of reasons, ten to a hundred times more effective than online ads. The other is to argue that advertisers are better at pricing online ads than offline ads.

In the days before the internet, advertisers made choices between billboards, radio and TV ads, the phone book and newspapers. All these ads work on an impression basis - it's very difficult to measure the effectiveness except via anecdote (I put up lots of ads and my sales increased) or by asking customers to bring the ad to you as a coupon. In comparison to the sort of detailed information on performance you can get from Google Ads, measuring the performance of newspaper advertising seems impossibly inexact.

Let's posit for a moment that the price of newspaper ads may have more to do with how much money a newspaper needs to earn to keep the presses running, rather than how effective they are at producing new business for advertisers. It certainly may make sense that the Eagle has to sell inserts for a CPM of $45 to continue producing the paper without suffering large losses. Why are advertisers willing to pay these prices without strong evidence that they give an effective yield? They may not have much choice - other options in a community where many customers are offline are also pay per impression and may be similarly expensive. The worry the local Price Chopper has is that if they don't produce an insert and the Big Y does, perhaps they lose their share of the local customerbase. Without good methods to track the effectiveness of the print ads, the Eagle's ability to sell ads may have more to do with comparable ad rates in other local newspapers or radio stations.

So what happens if the market rationalizes, if pay per performance advertising becomes a viable way to reach the majority of consumers who consume a particular publication? This may be what's happening to papers like the New York Times. As print circulation decreases and online readership increases, it seems like the newspaper could still afford to produce high quality journalism. But if online ads sell primarily on the basis of their effectiveness, and print ads sell for other historical, competitive and less rational factors, revenue could fall sharply as readership increases. There's certainly no shortage of speculation that the Times, like almost all newspapers, may be in trouble (though the Times's response to a particularly gloomy Atlantic article is a masterpiece.)

Whatever objections I have to the Berkshire Eagle's charming habit of making yesterday's weather the lead story in the day's paper, I don't want it to go out of business. And I really don't want the New York Times to go under. But I'm starting to worry about the irrationality of the model that's supported journalism for the past several decades.

I remember a conversation before an academic conference in 2006 with a New York Times reporter. In typical conference fashion, we'd been paired up, blogger and "real journalist" to discuss the future of journalism. Talking informally the night before, the reporter said, "I don[base ']t really understand how it is that Bloomingdales underwrites our Africa coverage, but as long as they're doing it, I'm not complaining."

I'm not either. But I am worried. What happens when Bloomingdales wises up and concludes that they'd save a small fortune by advertising online, targetting only to readers who actually live near their stores?

Here's my concern. If I'm right and print advertising costs are fundamentally irrational, then it's possible that the way we've built media in the United States can't survive a transition to a more rational market. That would be bad. Newspapers aren't just businesses - they serve a critical function in a democratic society, informing citizens so they can make intelligent voting decisions, lobby their elected representatives on issues of their concern and hold political and business powers accountable.

What if the idea that commercial enterprises should carry out the public interest function of journalism is built on a fundamentally broken model? What if advertising worked pretty well as a way of subsidizing public interest journalism only so long as advertisers didn't understand the effectiveness of their ads? Putting aside all the other reasons why commercial journalism may be flawed - the tendency of newspapers and television channels to seek readers by publishing "edutainment" rather than investigation, the worry that papers will hesitate to publish stories that might embarrass advertisers - what if ad supported journalism is only viable in a world where we radically overvalue the worth of ads?

That would be a bad thing. Seth Godin argues that there may be lots we're willing to either throw away from the local newspaper or have covered by new online providers, but we need to ensure that there's still a way to engage in serious investigative journalism: "I worry about the quality of a democracy when the the state government or the local government can do what it wants without intelligent coverage. I worry about the abuse of power when the only thing a corrupt official needs to worry about is the TV news. I worry about the quality of legislation when there isn't a passionate, unbiased reporter there to explain it to us."

Godin believes that we'll find another way to provide for this coverage. "Maybe it's a public good, a non profit function. Maybe a philanthropist puts up money for prizes. Maybe the Woodward and Bernstein of 2017 make so much money from breaking a story that it leads to a whole new generation of journalists."

Maybe. I wouldn't count on it. My friends who are engaged in online projects to conduct "difficult journalism" - the sort of investigative reporting Godin is talking about, as well as international coverage, are worried about revenue models. We get support from the foundation community, but foundations can't provide support forever, and all would like to know when we're going to be able to work without their support.

We're all looking at models that include some advertising support. What if the model that brough us Upton Sinclair and Woodward and Bernstein - impression advertising - can't bring us into the future because it's based on uneven distribution of information and bad math?

Friend and colleague Doc Searls has an important post that I wish I[base ']d read before writing this. He sees a similar set of problems with the rise of pay for performance advertising, and argues that systems where buyers find sellers - which he calls VRM (vendor relationship management) systems - will be an important force for supporting journalism in the future. Worth reading.