How To Pitch An Algorithm

The Pew Research Center’s (2015) annual “State of the Media” report always provides intriguing insights into US media consumption habits, and the 2015 report continues in this robust tradition. This year, we find out that contrary to reports of its imminent demise, the evening network news audience rose by 5 per cent and network morning news by 3 per cent. Local TV news was also up 3 per cent, while daily and Sunday newspaper readership declined by the same amount. The biggest shrinkage took place in prime time median cable viewership, which was off 8 per cent.

The biggest shock in this year’s report, however, was that, in 2014, 48 per cent of Americans with Internet access got their political news through Facebook. Even though these respondents also got some political news from an average of 4.8 other sources, the importance of Facebook as a news portal can hardly be overstated in light of one critical fact: the news that is served up to them is based on who their friends are and what the FB algorithm thinks (sic) they want to read, based on their past online behavior. We believe this has major ramifications for the way in which public relations people pitch stories to and through the traditional media, still rightly seen as an important gateway to public influence.

Practitioners of public relations focused on traditional media have, for decades, observed a few simple rules as they go about their business: know and understand an individual medium’s readers, viewers or listeners; assess the style and narrative flow of an outlet they are trying to reach; then craft a pitch that looks and sounds as much like an actual story that might appear in your target publication or network.

The emergence of search engines for the World Wide Web did not seem, at first likely to change this well-established practice. Early versions of what later evolved into search engines were little more than catalogues of documents on the Internet. “Archie” (archive without the “v”) was a simple list of files available on public anonymous File Transfer Protocol sites that debuted in 1990. The succeeding decade saw the development of increasingly richer tools such as Veronica, Jughead, W3 Catalog, JumpStation and Webcrawler. By the late 1990s, there was a plethora of search engines embedded in portals such as Yahoo, but it was Google’s adoption of Goto.com’s paid search strategy that enabled the “medium” to take off and become the core ingredient of the increasingly Web-mediated human experience.

As the era of search engine optimization boomed and digital readership became increasingly meaningful to traditional print news media, journalists and editors began to grapple with the significance of search rankings. A 2006 New York Times article by Lohr (2006) captured the new journalistic anxiety that accompanied the first attempts to adapt headlines to ensure that articles appeared high up in search results.


He quotes Michael Schudson, a visiting professor at the Columbia Graduate School of Journalism, who said: “My first thought is that reporters and editors have a job to do and they shouldn’t have to worry about what Google’s or Yahoo’s software thinks of their work.” Search engine optimization experts argued that the fear of search affecting journalistic integrity was overblown. In “Writing for the Machine: Hysteria among journalists”, consultant Spencer (2007) wrote: “the core philosophy here is for journalists to let go of their search engine ‘machine’ fears and simply embrace accuracy [. . .]”.

In the intervening decade, writing both headlines and copy to reflect search engine ranking algorithms has become the norm for most outlets and public relations professionals routinely consider search engine optimization to ensure that their press releases appear in RSS feeds and on news aggregation sites. It was perhaps inevitable that news organizations exposed to increasing margin pressures would begin automating the other end of the news process, the writing of news stories themselves. In 2014, the Associated Press announced that it would employ story-writing software (Yu, 2014) to “cover” the release of corporate earnings.

While a human editor would approve the copy, the stories themselves would be written by machines. Companies such as Automated Insights, Narrative Science and Wordsmith are now busily supplying content for AP’s NFL photo captions and the results for fantasy football match ups. According to Automated Insights’ chief executive officer, Robbie Allen, his company’s software is now also able to supply humor as an added option. The Los Angeles Times has also ventured into these waters with an algorithm that can churn out breaking news content. The byline on a March 17, 2014, story (Schwenke, 2014) about an earthquake outside Beverly Hills read as follows: “This information comes from the USGS Earthquake Notification Service and this post was created by an algorithm written by the author.”

If this trend continues, which we can reasonably presume it will, we will increasingly find ourselves in an oddly self-referential world in which “news” written by machines to be read by machines and served up to us by machines will former a greater and greater proportion of our information consumption, which leads us back to the question raised by Pew’s research. What tools should be used by communications professionals seeking to advocate for their clients or causes to influence the opinions and behavior of the desired target audiences?

If stories are being fed to Facebook members by algorithms that analyze their friends’ feeds and their own Web behavior, how can we successfully enter this walled robotic garden? We believe there are three key disciplines that will play an increasingly critical role in ensuring that our content gets served up preferentially by the Web’s digital butlers:

1. sharing strategy

2. visual information

3. narrative confirmation

Click on the link below for the complete article.

“How to pitch an algorithm” appeared in The Journal of Business Strategy, Vol. 36 Issue: 4, pp.56 – 59, and is reprinted with permission from Emerald Publishing Group Ltd.

There is 1 comment

Add yours
  1. Shameek

    Very insightful article. However, my only question being a computer scientist is why would you name this article – “How to pitch an algorithm”? In my opinion, the title is extremely misleading. It could have been as well been “How to pitch news content”. I was directed to the article because I wanted to read more about the methods or elements I could adopt when I try to pitch an algorithm/patent designed by me, to an interested buyer/investor. Your article does not discuss anything in that context.

Post a new comment