Optimizing the Link Economy

A Proposal by Jeff Jarvis
Tow-Knight Center for Entrepreneurial Journalism
City University of New York Graduate School of Journalism

(Download “Optimizing the Link Economy” as a PDF.)


I talk a great deal (too much, some would say) about the link economy as the essential architecture of media on the net. Now I am launching an ambitious research project at CUNY’s Tow-Knight Center for Entrepreneurial Journalism to bring facts and modeling to the discussion. The goals:

• to learn how to optimize the value of links—on both sides of a click;
• to explore a framework to enable content- and audience-creators to negotiate a mutually advantageous exchange of value—that is, a content marketplace appropriate for the net;
• to investigate alternative methods of content distribution and value exchange—asking whether it’s best for readers to come to content or whether there are alternative ways for content to travel to readers; and
• to investigate the link’s impact on the economics of the larger news ecosystem—e.g., how it can make news more efficient (this is where I say, “Do what you do best and link to the rest”).

To begin the discussion and the research, here is my treatise on the link economy. We’ll take this to media companies, aggregators, and others asking for data on their links and we’ll then use this to recruit researchers to analyze that data, test assumptions, and make recommendations.

(Note: For convenience, I use terms such as “audience” and “reader,” though they are presocial, one-way-media words. As Jay Rosen would say, these are the “people formerly known as the audience.” Stipulated. “Reader” is also a Gutenberg-era word (though a long era it has been) portraying media as text. I do not mean to exclude other media, nor do I want to bother adding “viewer,” “listener,” and “user” at every reference. Rather than inventing new and inevitably grating terms for these broader constituencies with new roles, please be generous in your interpretation of “reader” and “audience.” I use them for the sake of brevity and simplicity.

The hypothesis

There are two creations of value in media online: the creation of content and the creation of a public—an audience*—for that content. Online, content with no links to it has no value because it has no audience. It gains value as it gains links. Thus something of worth is created on each side of a click.


What is that value? Well, of course, that is impossible to calculate precisely; there are too many variables. What we will ask instead is how to optimize the value of the link from various perspectives.

From the link originator’s perspective:
• How does a link send more and better audience to a source of content? There are many was to define “better audience:” loyal readers or readers whom advertisers value, who buy subscriptions, who make purchases, who recommend and distribute content, who contribute effort or comment.
• How do we identify those readers on either side of the link? How do we communicate information about a reader’s value (e.g., demographic or interest data) to the link recipient?
• How do we identify the original sources that we believe deserve links (because they invest in quality content) more than the down-wire repeaters and remixers of information?
• Most important, what links do readers value most? What is the most effective form of a link: a brief headline, a provocative pitch, a rich description, a sample of what is on the other side, information about the source, recommendations of the source, data about the popularity of the link or source, relevance to the link originator’s content, currency…?

From the link recipient’s perspective:
• How does the beneficiary of the link get the greatest value on its side of a click? How does it best establish a relationship with a reader and extract value from it?
• What would help the link recipient do that: sending better qualified audience to more relevant content; data to enable the recipient to realize more value in the relationship; or new ad sales and monetization capabilities (e.g., someone to sell national advertising to national audience while a local source sells its local advertising)?
• How can the recipient gauge and communicate a link sender’s performance and how could the recipient help the sender improve that performance?

An alternative view

From the reader’s perspective: There is another way to look at this question. We now make readers come to content: to our publications, shows, sites, and apps. Might it also make sense for the content to go to the readers, to become embeddable and spreadable like YouTube videos that travel the web with brand, monetization, analytics, and return links attached? Call this the peanut-butter strategy. Or the delinked link economy. What if there were a new content marketplace that enabled such an exchange to occur? What would that do to the relationship of the linker and the linkee, the aggregator and aggregated?

What if, for example, rather than complaining about The Huffington Post’s long, summary links to its articles, a newspaper could take advantage of interest in its content by placing full articles and photos—with its own advertising—on Huffington Post? Would the paper receive more readership and more value? For the sake of discussion, let’s say it would. Then the paper might encourage HuffPo to embed more of its content. Rather than charging Huffington Post for use of content as in the old syndication model, the paper might reward and motivate Huffington for embedding content with a share of its incremental revenue. Huffington Post then becomes a distributor more than an aggregator—of the paper’s and others’ content, maximizing the value of its audience. Journal Register is building a common content distribution infrastructure for its sites that could support such a model. One could imagine other services with large audiences and high engagement—Facebook, for example—taking on this distribution role.

I was shocked when I first used the Flipboard app on the iPad because content is there without ads or other monetization, without navigation and links to the content creator’s site, without analytics the creator can capture. How much better it would be for the content creator if there were a means to share full content in the app with those things attached. One outcome of this research could be a proposal for such a standard that might allow a news site’s article to appear on Flipboard with ads, branding, navigation, and analytics.

In these two examples, content travels to readers in a new form of reverse syndication. Where the content is read becomes less important than the fact that it is read, the knowledge of who reads it, and the ability to exploit that knowledge. When both parties have an agreement to share revenue and business rules to govern that exchange, then who sells the advertising becomes less an issue than who can get the highest return for both parties. Ads could be sold by the content creator, the distributor, or a third party (a network or a sales agent such as Google).

I don’t suggest this model as necessarily a substitute for the current structure of media sites as distributors of their own content, acting as magnets for audience. But it is a useful illustration that leads us to examine and question some of the assumptions behind the current structure of online media.

What stops the reverse syndication model I describe from happening today? At least two issues: First is the lack of a means to determine, negotiate, and exchange value—a marketplace. Second is measurement, or what I’ll call the total-audience myth: We remain tied to an old-media model of measuring all the audience that comes to a brand because that’s what legacy media forms required: The advertiser was charged based on total audience thanks to the convenient myth that every reader and every viewer saw every ad. There was no way to verify who really saw each ad, as there now is online. Newspapers were motivated to gather the largest total audience. Advertisers calculated return on investment based on total audience, so that is how they shopped for media, starting with the largest properties. For the most part, ad agencies still buy that way online, even though on the web they pay only for the ads that readers see (or better yet, click on).

There is no scarcity of advertising availabilities online and so it is a wonder that advertisers do not more aggressively exploit that opportunity to put constant pressure on ad pricing. I am often asked why, if online advertising is more measurable, targetable, effective, and efficient, it costs so much less than advertising in legacy meeting. The answer, in a word, is abundance. Old media controlled and sold a scarcity of space or time and thus held pricing power. Online media face no limit of competitors, which will only put downward pressure on advertising sold the old way. Advertising must be sold now – as Google sells it – on the basis of performance and return on investment.

It makes little sense anymore for advertisers to concentrate their buying solely on the largest sites when they could take advantage of the abundance the net creates by putting together more targeted, less expensive networks across multiple sites. (Building those networks takes time, effort, and expense. One opportunity we will examine is how these smaller, more specialized competitors could gain business by creating networks of their own or providing more useful data about their audiences to networks built by media buyers.)

Similarly, we know that it no longer makes sense for media to attract the largest possible audience if it cannot monetize all of that audience. See the local news site that considers national traffic an expense because it eats up bandwidth and is of no value to local advertisers. See also the newspaper that killed its stock tables, saving $1 million a year in paper and ink while losing a dozen subscribers. That means the paper had been spending $83,000 per year to hold onto those subscribers, who clearly could not return that value. We get what we measure and we are measuring the wrong metrics.

Considering the value of readers and relationships

Rather than concentrating on total audience, we should concentrate on the net future value of each reader. Where does that value reside? That question raises a fundamental strategic—and religious—issue: We in news and media keep saying that our content has value. Well, yes; no one will disagree. But we need to ask whether the greater value resides in the content or in the relationships and data it can spawn. Yes, the content has value, but how best do we extract that value?

Over lunch recently a media executive repeated the accepted wisdom that “our content has value.” That often leads next to the contention that we “should be paid for it,” though I counter that “should” is never the basis of a business model. In news, of course, we have always extracted more value for our work through selling our audiences to advertisers than selling our content to audiences. Why would that change today?

This executive also complained that digital companies, such as Google and Facebook, don’t value our content. But look at this new media ecosystem from the perspective of Facebook, a company that by some reckoning could be valued at as much as $100 billion by the time it goes public within a year. What does Facebook itself value? Relationships. Data. Relevance.

As for content, Facebook doesn’t so much refuse to value it, as my media friend implied, but instead finds value in a much more expansive view of content. It finds worth in all that apparently useless blathering we do in what Facebook calls, to journalists’ derision, its members’ “News Feeds.” That’s not news, the news people say; news is what we make. That may have been the case in a scarcity-based content economy, when there was room for only so much news in the world’s publications and airwaves. Now content—like advertising—is abundant. The incumbent content companies are having trouble taking advantage of that growth because their definition of content remains limited and their models based on controlling scarcity. Facebook, like Google, sees content everywhere, made by everyone, and each in its own way is better than legacy content companies at finding value in it. Each uses content to gain more signals about users and to use that data to target content, services, and advertising.

My lunch companion said that media companies’ content is the “steel” that makes Google’s “cars.” That metaphor still assumes that content is a scarce, consumable, and perishable commodity. Digital companies’ ability to make money on the back any content—Facebook enables the creation of it; Google organizes it—irks the content makers. This is why Rupert Murdoch and his News Corp. lieutenants (in a list curated by Arianna Huffington) accuse Google and its ilk of being “parasites,” “content kleptomaniacs,” “vampires,” and “tech tapeworms in the intestines of the Internets” who “steal all our copyright.”

There are two problems with the Murdoch worldview: First, according to my thesis of the link economy, Google, Huffington Post, curators, aggregators, bloggers, and readers linking via Facebook and Twitter do not steal value but instead add value when they direct readers to content. In response to News Corp.’s accusations and epithets, Google Executive Chairman Eric Schmidt said in Murdoch’s own Wall Street Journal in December 2009 that Google causes 4 billion clicks a month to news publishers, a quarter of that from aggregator Google News.

In an apples-to-pineapples comparison, only a few months later, Bit.ly, the leading URL-shortener used in Twitter, passed that 4 billion mark and a year later it doubled that (though not all that goes to news sites). There we see the rising power of the peer’s recommendation, the human link. In early 2011, the Pew Research Center’s Project for Excellence in Journalism confirmed that social services were driving higher proportions of traffic to news sites, with Facebook coming in second or third in the list of referrers to five of the top 25 news sites.

The second issue with the Murdoch view of links is that it fails to take account of the new ways that digital companies mine value in content, links, and relationships. For them, content is not a product to sell but is more a device to generate information about users to increase their value. Content is a signal generator that reveals interests, needs, sometimes location, and more. Facebook can find out that you are a fan of Green Day if you read articles about it but also if you write about it or your friends are fans or you listen to or recommend its music. Then Facebook wants to sell you a ticket to the next Green Day concert near you (and Facebook knows where you are). In this example, content takes many forms—an article, a conversation, a song—and monetization comes not from advertising but from commerce. Does Facebook need a publisher’s article to make these economics work? Is it the steel without which there can be no car? Hardly.

A more extreme example: In 2010, researchers used a set of keywords to track aggregate moods in Twitter messages and found they could predict daily ups and downs in the Dow Jones Industrial Average with up to 87.6 percent accuracy. A hedge fund now uses the formula in partnership with one of the scientists. The content—very broadly defined—created by millions of Twitter users produces value, if you know how to look for it.

In our research, we will need to catalogue such additional sources of worth and revenue. For part of the lesson to content creators and link recipients should be that there are more ways to recognize value than the traditional way of selling audiences to advertisers. At the e-G8 conference in Paris in May 2011, Zuckerberg bragged that Zynga, built atop Facebook’s open platform, had just past game champion Electronic Arts in market capitalization. He said Zynga succeeded because it understood not only games but also people and relationships. He suggested that the next winners in music, for example, would similarly understand both (see: Lady Gaga). How will the similarly savvy news company succeed?

I’m not suggesting that editors call the people formerly known as the audience little monsters and don bodacious bustier to earn a buck. But I do believe we must challenge our every assumption about the role of content and its creators in a new media economy. Media’s role was to make and distribute content because it controlled the means of both. Now they do not. The former audience can make content and media’s role may be to support them in that with tools, platforms, aggregation, curation, promotion, training. The former audience has also taken over the role of distributor when they link, recommend, discuss, and embed content and so the question for media is how to take full advantage of that. Where do the former content controllers fit into this new ecosystem? How do we add and extract value?

The simple question—how do we increase the number and value of links and clicks for media—raises these larger questions. This research can hardly answer them all but perhaps it can inspire new ways to see value and new structures and methods to realize it.

Next steps

This document is intended to be the basis of discussion around how to engage in research on the link economy. We will seek help and guidance from many. These are possible next steps:

• We will gather data from media sites, aggregators, ad networks and agencies, and any other sources that can help us get a picture of the behavior of links and clicks: how much traffic does one site send to another and what happens to that audience. We will attempt to categorize links (e.g., headline vs. promotion vs. sample) to compare their relative performance. We will begin to create a taxonomy for the value of links (e.g., ones that cause more traffic or send audience that behaves in more valuable ways). All data will be anonymized as to users and, when requested, as to sites.

• We will seek partners to run A/B tests to compare the performance of links in controlled circumstances (e.g., for the same story, does a headline or a headline and sample send more traffic and what becomes of that traffic on the other side?).

• We will seek partners to run tests of reverse syndication.

• We will work with partners to model link behaviors. See the excellent work of Boston University Prof. Chris Dellarocas examining the mutual benefits in a link economy.

• We will hold workshops and meetings to discuss goals, data, and preliminary findings and to ascertain issues and opportunities with link originators and recipients.

• We will publish best practices related to links: how to send more and better traffic and how to better recognize value from receiving links and establishing relationships with audiences.

• We will publish recommendations on how to optimize the performance and value of links, including how various parties may work together to share audience, data about them, and value. This may include suggestions to build new standards, platforms, or possibly enterprises to facilitate an enhanced marketplace.

(Download “Optimizing the Link Economy” as a PDF.)