As long as digital publishing flattens the business value of content, there is little hope to improve the economics of news publishing. The problem needs to be addressed.
The biggest failure of digital news platforms is their inability to assign a different economic value depending on a story’s type and quality.
On a news site or an app, advertising space is sold for the same price whether it’s a news wrap-up quickly put together by a junior editor, or a 10,000-word piece that required a whole team of writers, editors, and fact-checkers.
The only element impacting the CPM (cost per thousand impressions) is the ad module placement. An ad located on the homepage of a section will be many times more expensive than one buried elsewhere on the site; similarly, being at the top of the screen pays way better than being at the bottom.
But the economic yield has nothing to do with the size, depth, uniqueness of a piece, or its authorship. It applies for text as well as for video: dash-cam footage of a spectacular crash in Russia (an unfathomable topic) is likely to carry the same CPM as an in-house documentary.
This promotional video provided by SpaceX seen on Bloomberg carries the same ad rate as numerous segments produced by the Bloomberg video team. (In passing, as another proof of the news industry’s decay, Bloomberg is showing the very same ads, over and over and in an unskippable format, on every video capsule).
By contrast, the TV industry has long developed robust dynamic pricing strategies. Using algorithms devised by yield management specialists, sales houses have created sophisticated models that charge advertisers dozens or sometimes hundred times more for an ad played during high audience programming such as sport events, versus clips shown before yet another rerun at 2am on a dull summer night.
Being able to connect the value of a content (a news story, a video, a multimedia package) to the price charged for it could greatly benefit publishers.
The difficulty lies in assessing the dual notion of quality for a news story that is both static and dynamic.
The value of a chunk of mozzarella can be captured with a small subset of objective elements: texture, odor, taste, origin, maybe a label of quality. These are static intangibles—elements that will condition its economic value. By a contrast, a news element carries a dynamic value that is likely to significantly change over time.
Let me take the example of a fictitious (and tragic) news cycle.
One winter day, a Boeing 787 Dreamliner inbound to Chicago O’Hare crashes in a freezing Illinois cornfield. 200 people are dead. This is a terrible human tragedy, but also a serious problem for Boeing. The company launched its new aircraft just a few years ago after three years of repeated delays due (among other things) to manufacturing problems involving a never-used-before carbon composite material (more on this in a moment).
The first part of this tragic developing story involves endless updates on the aftermath of the crash: mourning the victims, detailing the circumstances and causes (largely unknown in the first 24 hours of the incident), and providing speculative comments from legions of experts hastily lifted from oblivion.
Important as they might be, these are typical commodity news items: every outlet across the country gets the same stream roughly at the same time. Quality (taken in the traditional journalistic sense) is so-so because rumors tend to flare without much thoughtful fact-checking; everyone is in the same boat. The economic value is near zero. It can even be nil altogether as some news organizations chose to forgo advertising in the case of tragic events—advertisers don’t like that sort of juxtaposition anyway.
As the news cycle unfolds, things change. At least in theory.
In every large newsroom of the country, reporters on the transportation beat will move heaven and earth to squeeze their sources in the airlines sector, especially at the National Transportation Safety Board in charge of the post-crash investigation. At this stage, the notion of exclusivity sets in: if a reporter gets her hands on a flight-recorder transcript, the story will score high in terms of economic value. In theory. In fact, due to the structure of a traditional website and the way ads are sold, this scoop is unlikely to carry a higher value than yet another third-hand Kim-Kardashian-robbed-in-Paris “report.”
This unfortunate trend has been reinforced over recent years because the share of sales done through automated market places is on the rise: according to eMarketer, the share of programmatic ad spending will reach 73% of the total US market with a year-to-year growth of 44%. And none of the machines that assign an economic value to journalistic work are able to see the difference between a conspiracy theory and an investigation!
You get the point: the market doesn’t reward exclusivity, nor traditional journalistic legwork (the painstaking but crucial craft of cultivating sources).
Back to our fictional example. At some point, the SeattlePI news team following the Dreamliner crash gets a critical hint. According to an anonymous NTSB source, the plane might have suffered some kind of structural failure as a result of a severe turbulence encountered during its descent to O’Hare. This is big. The sizable and documented story that ensues draws a large audience. But, yet again, this high journalistic value barely moves the economic needle: ads associated to this breaking news—pardon the pun—won’t budge, especially since the story will be picked-up (with or without proper credit) in a matter of minutes: When CNN echoes the SeattlePI scoop, its version gets ten times the traffic of the original piece.
Some might object that a great journalistic piece might indeed bring more money to the publisher because it draws a larger audience. In theory, yes. However, hard news performs poorly compared to cheap feature stories. For example, last week on Business Insider, a story about the new Facebook classified service said to sell drugs, animals, and adult services raked up more than 800,000 views; the classic “10 must-have travel tech accessories” got more than 300,000 views. At the same time, very few tech or finance stories in BI reached the 10,000 mark that week.
Now, let’s use a different angle.
Three months before the Dreamliner crash, an obscure university professor specializing in composite structures wrote a piece about risks arising from the use of carbon fiber in large structures such as airframes or wings.
(Note that, and this is quite important for the rest of my demonstration, there are plenty of highly-specialized sites manned by monkish experts like this one, Atomic Delights, which covers the intricacies of manufacturing and metallurgy—it was pointed out in a recent piece by my co-writer Jean-Louis who shares my passion for tech-porn sites.)
At the time of its publication on the confidential site, the piece about the hazard of composite materials went totally unnoticed. Albeit technically irreproachable, it had little journalistic value…. Until another reporter from the SeattlePI unearths and republishes most of it (again, with proper credit and link.) The original author got plenty of views. Assuming the site had a sponsor, it made some extra money. In that case, context and time had increased the value of an in-depth work.
Three important points to conclude:
- While journalistic value or news-worthiness varies widely, economic value remains desperately flat.
- The web is filled with specialized and unique contents that carry high journalistic value, but they are barely detectable.
- In order to reconcile journalistic and economic/commercial values, the way advertising is sold against content must be rethought from the ground up.
There are several avenues to address these three challenges:
Every story coming out a Content Management System (CMS) must carry a set of precise “signals” (tags) for quality: an investigative piece as opposed to a wrap-up; an in-depth profile vs. a short capsule, etc. These signals (or dedicated tags) should also reflect the number of people who worked on the piece and be adequately machine-readable by an ad server in order to spot unusual value and raise the formats’ prices.
These signals must also be detectable by secondary distribution platforms such as search engines, social networks, and aggregators that will grant a higher ranking to the flagged content.
The question of the “undetected quality,” such as the carbon composite specialist operating under the radar, is more complex. We can hope third party companies will come up with solutions: in the same way small ad networks aggregate small outlets and sell them as a bundle, they should consider applying the same tagging system to flag (and monetize) quality.
I’m aware of hurdles such as building a foolproof system, scaling it to cover scores of topics to create critical mass and, above all, making a true standard widely adopted. A long road indeed.
Inevitably, the necessity of training these algorithms will arise. It involves machine learning and neural networks that will process millions of stories—first to understand, and then mimic how human editors evaluate editorial quality, long term value, and uniqueness in a news cycle.
This is the project I’ll be working on during my year at the JS Knight Journalism Fellowship at Stanford. Stay tuned. (And subscribe to the Monday Note.)
This post originally appeared at Monday Note.