Lenfest Institute experimentation grant applications

So if you’re reading this, you likely know about the Lenfest Institute, a first-of-its-kind entity working on the wicked problem of local news. The Institute opened up its grant applications for the first time this summer and I submitted a few applications to the experimentation fund.

Sharing the abstracts in case other people have prior art or insights to provide. The ideas are cheap! Making them happen is the hard part 😄. Only one I’ve started developing so far is the POUND prototype but I’m confident these are all technically feasible.

 

Optimization of on-site widgets through real-time programmatic negotiation between advertising, subscription, and audience development systems.

A classic hard problem in news design is choosing which widgets to place on a page. Current approaches to this problem are generally static – the same layout gets served to all users.

Yet we know not all users are created equal, nor are all pageviews.

Facebook has trained AI bots to negotiate using its Messenger service. Advancements in open source machine learning have made this practical and cost-effective. I propose to use similar methodology to create a first-party system wherein parties representing advertising, subscription, and audience development can negotiate in real-time for placement on the article page.

Real-time bidding/programmatic advertising are well known in the publishing industry at this point. But dynamic layout projects are generally limited to academic efforts.

Links to other papers on subject…
http://bit.ly/2t9TkmX
http://bit.ly/2siv3uP

Measurement is itself a hard part of this project.

To make the negotiation algorithms practical, we must also establish functions that approximate the value of actions including but not limited to the following:

  • Newsletter signup
  • Account creation
  • Social media follow
  • Comment
  • Event registration

The key insight here is that user interactions outside of advertising have corporate value; by working to de-humanize these decisions we can realize more rational and optimal outcomes.

 

Create a new kind of mobile interactive advertising format that’s sneakily good – micro surveys that serve as brand marketing, lead generation, and first party data collection.

NPR has shown that game mechanics, like asking users a question, can significantly boost conversion rates on donation forms. By using simple interactions – just a couple buttons to start – we will blur the line between content and creative and draw users into the brand experience.

This interaction format had been pioneered previously by Luke Wroblewski’s startup Polar (acquired by Google), and Google surveys remains a popular revenue stream for publishers. Surveys are also an ad product at the Washington Post and other digital media companies.

New aspects to this iteration of the idea include a wizard for quickly building out creatives; a first party tool for capturing and viewing the leads; and the ability to send survey responses along to a data management platform (DMP).

As an advertising product, the experience will need to be as enticing as possible, which we will achieve through human centered design and user experience engineering. As a first party advertising product the tool should be able to circumvent a greater number of ad blockers than typical display ads.

The closest in-market competitor is the Outline, which has developed original ad formats that are reportedly converting at above-average rates. This project would be different by focusing exclusively on the ad format, and by further exploring the links to DMP and future targeting efforts.

 

Structured business model experimentation around archival content.

A chief asset of legacy media companies is archival content. Digital replica products like Newspapers.com can provide a great revenue stream, but is it the only one?

Business models like events, podcasts, gifts, and research services can be seen in the existing “history” market both locally and nationally, but few metro newspapers have begun experimenting with their own products in this space. An exception is the New York Times, which has an extensive online store.

“Who controls the past controls the future,” said George Orwell once. As metro newspapers develop product lines around historical content, they will reinforce brand identity as being not just reliable information providers, but as critical institutions that have helped their cities grow.

This idea is especially timely given the link between nostalgia and political turmoil.

There’s a ton of goodwill for legacy media companies, specifically around historical content, that could be tapped in the short-term to help ensure long-term survival.

 

Recreate the BuzzFeed technology, POUND, in an open source project.

POUND is the Process for Optimizing and Understanding Network Diffusion – BuzzFeed publisher Dao Nguyen revealed its existence an April 2015 blog post discussing the massive traffic success of “The Dress.”

POUND visualizations and the underlying data analysis help newsrooms understand how content bounces and spreads between networks.

Organizations including Mic, the Washington Post, and the Marshall Project have implemented their own versions of this tracking approach, but this would be the first open source and generic solution with community support.

Outside of the media outlets mentioned earlier, I am not aware of other tools that track the *spread* of content through social media platforms, although there are plenty of tools that track the simpler measure of *popularity* across different social media platforms.
Posted Jul. 3 2017, 7:37 am by davis