That’s why I was happy to see Jeff Reifman of NewsCloud open up about how their recently concluded Knight Foundation grant didn’t deliver what they’d have hoped. Reifman’s final report discusses the difficulties their project — which aimed to partner with news organizations to build online communities around Facebook — faced. Some are familiar complaints: difficulties getting buy-in from large news outlets; struggles to interact with their technology side; and problems dealing with a shifting underlying platform (in this case, Facebook).
Reifman is writing from his perspective, obviously, and the news orgs he worked with would likely have a different take — but he does try to note some of the complaints his partners raised here too. For others who’ve tried to partner with news organizations — or who’ve been on the other side — anything in here look familiar? If you’re starting out on a project of this sort, I’d recommend giving NewsCloud’s report a read to get one take on some of the problems you might face. (For a pdf version of this report, go here.)
In December 2009, NewsCloud received a $249,942 grant from the Knight Foundation to work with twelve news organizations to help them launch interactive community sites based on our open source Facebook technology. In its final report form, the foundation asks, “What headline best summarizes the single most important outcome of your work?” Sadly, “Social Media Technology Fails to Save News Organizations from Themselves” is my response.
We’ve been told that the lessons we’ve learned about how to work with media organizations in the age of social media are the most valuable outcome of this project. It’s for this reason that we think it’s important to be open about our successes and failures.
Our primary belief is that news organizations have evolved to favor a culture of one-way publishing. In other words, they gather the news and publish it to their paper and website. They encourage readers to interact by sharing stories and adding comments, but that’s often the horizon of their vision and technical capacity. They often view social media as a tactic for driving page views, not as a fundamentally new way of engaging their audience. The NewsCloud platform has a larger vision for news organizations’ role and social media.
Our platform is designed to help news organizations host vibrant communities for their readers to meet each other and engage with published content and also to generate and interact with their own content. Our integration with the Facebook platform helps readers see each other’s names and faces and fosters a more civil environment than the web’s typical anonymous content.
Some of the features we offer include end user blogs, discussion forums, questions and answers, a directory of resources, events, multimedia galleries, idea gathering and even a predictions game. We recently added a classifieds system, which allows for lending of physical goods within communities. You can see a variety of these features at our Seattle prototype site, The Needle (http://theneedle.us).
A big part of our vision is for journalists to integrate the interactive features of our application into their storytelling, expanding the space for reader participation. For example, The Washington Post’s advice columnist Carolyn Hax is creating resource directories of links relevant to her daily columns. We foresee reporters setting up idea gathering boards for ongoing series on civic issues or setting up prediction topics and questions on political stories, e.g. Will President Obama be re-elected?
Our model aims to increase user engagement and brand loyalty, which we believe will ultimately lead to increased traffic and revenue.
It’s useful to note that in our earlier Knight Foundation-funded research study on the engagement of young people in news, we found that our Facebook technology laid the groundwork for this vision: 1) It created a platform for the audience to become familiar with each other’s names and faces and drove return visits to see what these new people were posting and 2) It created a ladder of engagement from lightweight to more involved which allowed the audience to connect with the community in ways that fit their initial level of interest. This held the potential for moving them along the ladder towards deeper engagement over time. See http://newscloud.com/research
Working with fifteen carefully screened organizations, we designed and activated sites for eleven media partners. Five of these sites remain active: Advice columnist Carolyn Hax of The Washington Post, The Boston Globe, The Charlotte Observer, Minnesota Public Radio’s MinnEcon and Baristanet. By mutual agreement, KPCC Radio of Southern California turned off their site in January 2011. Five partners walked away from their sites without launching: A Cuba site community of The Miami Herald, The State Press of Arizona State University, an oil spill community site for the gulf coast by the Society for Environmental Journalists and two Gannett-owned sites, The Detroit Free Press and WKYC NBC 3 of Ohio.
Of the five active sites, only a few have been actively promoted – and generally, only occasionally. For the most part, we feel that the NewsCloud platform has yet to be thoroughly experimented with in a way that will both provide the feedback needed for improving its feature set and identifying best practices. In a word, the jury on this kind of community technology is still out – it remains a vision of community building that most of our partners say they believe in, but rarely put the resources required into turning the vision into reality.
We acknowledge that our partners faced many technical challenges working with our small grant funded team, tight on resources, building an emerging community technology. The biggest technical barriers dealt with design aesthetics and usability, feature limitations and bugs. These gaps made it difficult for our media contacts to wholly embrace our technology consistently and slowed adoption and evangelization in their organizations. However, these kinds of challenges are fairly typical of all software projects.
In an age of social media and a plethora of new technology tools, it’s useful for media organizations to become savvy working with software developers and actively participating in the path towards higher quality inherent in these kinds of projects. Some of our partners were more experienced with this process than others.
It’s possible that we could have screened partners more carefully and asked that they make more explicit contractually binding commitments. However, given my experience on this grant, it would have been extremely difficult to recruit any partners if this were the case.
To be perfectly clear, while we’re very satisfied with the technical progress on our project, we are not satisfied with the results from our work with media organizations. Most of our partner sites remain idle awaiting deeper promotion.
Our goal for this post-mortem is to transparently highlight the barriers that we encountered for everyone to benefit from; we are not trying to single out partners nor point fingers. Furthermore, we acknowledge that this represents our biased point of view. In the next section, we’ll highlight how the points of view of our partners vary.
Here are some of the more significant challenges we’ve encountered:
Our belief is that by far the biggest limitation to progress and growth on our partners’ sites was a lack of promotion and visibility. While The Boston Globe and The Charlotte Observer briefly experimented with posting links to application content on their home page, none of the partners regularly promoted their sites in front of their web audiences. None of the partners provided clear, meaningful links from their home page to the application or its features.
Just as the sound of a tree falling in the forest is lost if no one is there to hear it – similarly, a community Facebook application is useless if no one knows it’s there or why to visit. Many of our partners felt that these sites would grow on their own accord without significant promotion. It’s been a year of magical thinking.
We regularly tried to educate partners on how to promote these kinds of applications and have written up some basic guidelines. For the most part, our advice has been ignored.
In general, our partners would try to promote their community site by posting links on their Facebook page. However, the audience of their Facebook page was only a tiny fraction of the audience of their website. So, while they did indeed see a typical 2 percent response rate from their Facebook page audience, this was not enough to seed their community. They were essentially expecting a fraction of a fractional audience would be sufficient to launch a new community site.
Our radio partner KPCC said they used a significant number of on air promotional spots. However, it’s difficult to get listeners to go online to a Facebook application from a radio spot. KPCC did not link to their community site from their website until very late in the project, and then only from the very bottom of a secondary social media page. Also, the KPCC site concept was based on a group of volunteer bloggers who created daily content. In their post mortem report, KPCC reported struggling with deciding whether to promote the user generated content from their website as well as technical challenges with the content management system. This decision likely doomed its application.
As early partners left their sites un-promoted and idle, later partners drew conclusions about the platform’s efficacy and seemed to slow their efforts and lessen their own commitment.
We’ve also come to see the limitations in partnerships when a media partner is given something for free. We put a lot of work into screening partners so that they knew what to expect and that we knew what resources they would commit. However, in most cases, the initial enthusiasm of media partners to work with a Knight Foundation grantee project did not translate into long-term organizational adoption.
In the case of our grant, media partners essentially received a “gift” of our open source Facebook community site, fully hosted and supported as well as a team responsive to their feedback and feature requests. But, because the organization wasn’t paying for their site, their commitment to the project seemed to wane over time. Most of our partners did not follow through with their commitments to promote and participate in the project.
Yet, if we had required partners to pay to participate in the project, it’s unlikely any would have signed up. Because of the financial and organizational strains in the media industry, almost none of our partners would have contributed matching dollars to our project. However, some partners dedicated moderate staff time – often beyond our appointed contact.
At the beginning of our grant, it took us several months to prepare the Ruby on Rails source code to be able to launch the first partner. But, in general, many of our partners took a significantly extended timeframe to launch.
It generally took NewsCloud only two to three weeks to design, set up and prepare a site for launch. However, our most active partners still took between three to five months to launch. This was often due to bureaucratic challenges with legal, IT or other organizational issues (addressed below). Partners that withdrew often took between three to nine months to make the decision to exit.
Our grant period was eighteen months. Having partners take up to 5 months to launch took significant time away from the length of time a site would be live for audience feedback. Similarly, the extended timeframe prior to withdrawal used up spots that other news organizations might have benefited from.
Overall, the inertia required to instill institutional expertise in social media and create behavioral changes in these organizations seemed insurmountable at times. The sites we built for our media partners often remained in a silo with our primary contact and there was rarely crossover awareness or adoption of the site within the overall organization.
Many of our media contacts we trained did not have the forum, influence or charter within their organizations to share what they learned about our application platform with their peers. This made it especially hard to connect with our partners’ reporters to help them think about the potential of using our interactive features to engage readers in the storytelling process (described later in the Looking Ahead section).
Surprisingly, we had an enormously difficult time explaining to the organizations that they wholly owned their Knight-subsidized sites. They could both fully realize any advertising revenue generated by the applications and they could also track and include page views from the application in their overall activity statistics that they use for ad sales. Despite this, many organizations stalled at providing the information we needed to integrate advertising and statistics. We are still routinely told by partners that their organizations are hesitant to promote or link to their application for fear of diverting page views from their website. We have not been successfully able to convince the institutions that they wholly own these community sites. Nor, have we been able to instill in them our vision of the potential upside and revenue generating potential of a vibrant active social media community site.
The NewsCloud platform allows partners to layout their home page with a variety of different widgets. Almost without fail, our news partners would lay out their home pages to make their community sites look like identical copies of their news websites. This approach fails for a number of reasons.
Firstly, a community site is different than a news site – it should feature and highlight the best of user-generated content. Second, if the community site looks like a carbon copy of the news website, readers won’t understand the purpose of the community site and how they can participate.
Gradually, we’ve worked with partners to configure their community sites more effectively. The MinnEcon site is a good example of this. See http://minnecon.mprnews.org.
We repeatedly ran into bureaucratic delays that greatly hindered, hampered and doomed partnerships with some of our major media partners. Mainly, bureaucracy arrived in two areas: legal and IT.
The legal departments we worked with struggled with how to manage the liability issues of working with a small grantee on open source software. Despite the fact that these companies were getting our help installing and managing our open source software for free, they wanted us to assume all of the liability risks that they faced as billion dollar media companies. Obviously, our grant did not provide resources for this.
Some legal departments struggled with the very idea of using open source software despite it being widely used in other areas of their website.
IT departments struggled with how to allow an external project to proceed without their oversight. In many cases, it became impossible to get required information from internal IT departments. Simple things like analytics codes or advertising rotation scripts were just never delivered. In other cases, it would take weeks to get a sub-domain address mapped to our server for running our partner’s community site. Getting workable embed code from third party services like the awful and proprietary Brightcove video player proved insurmountable for several partners.
At the beginning of our project, some of our administration tools worked best with the Google Chrome browser. Some of our partners were prohibited from downloading the browser on their computer. One partner used Windows over Citrix for security (an operating system run over a network) and was unable to complete drag and drop operations required to feature content with our administrative tools.
Another partner was prohibited from using instant messaging in the newsroom, making it more difficult to work with us.
In general, our organizational contacts were already overburdened with a variety of responsibilities and focusing on promotion of their community site quickly fell to the bottom of their queue. Repeatedly through our project, numerous partners told us that they would return to focus on the application after this distraction or that distraction.
In some cases, staff was furloughed at different times during the project interrupting or delaying our progress. In some cases, budget and staff cuts were made in the middle of our partnerships. This seemed to contribute to some partners walking away from the project.
Said one partner: “In the midst of all this, multiple staff changes and cutbacks within our department left with left staff and more work than ever to do – meaning we had to prioritize and make hard choices about how to spend time and resources. Thus the decision to discontinue our part in the project.”
We also had situations of key advocates in the organization leaving their company during the middle of our project. More recently, one of our contacts is leaving their company and it is unclear who will take over our partnership and who will manage their site community.
Most partners were concerned about the cost of hosting and managing the application after the grant ended. At least one partner seemed to lose interest after learning that it would cost approximately $4,000 per year to run the application. We always find it disappointing that news organizations will invest full time staff in a project but balk at reasonable technology costs, tiny in comparison to staff salaries and marketing investments.
Our grant provided approximately 1¼ software development resources. For a complex platform like ours, this is a relatively small amount.
We took the approach of working most actively with the partners who worked the most actively to test the site and provide feedback. The Boston Globe and KPCC were very active in this regard.
KPCC asked us to deepen our blogging capabilities. The Boston Globe worked with us to improve the level of polish in the application and its features.
The partners involved earlier in the grant suffered a bit from the challenges with the initial releases. These were stymied with a limited feature set as well as bugs.
We couldn’t make our platform do everything, nor could we make individual features as robust as some partners wanted them. For example, our blogging feature is not as sophisticated as WordPress, and this disappointed the KPCC volunteers. Similarly, the design decisions we made around image handling in the application complicated the layouts used by The Washington Post.
We always did our best to address the major barriers faced by our partners but at some level our overall resources limited us. In some cases, our partners accepted the shortcomings of the tool and the delays in meeting their needs. In other cases, these problems contributed to the lack of adoption, marketing and withdrawal by partners.
Working with the ever-changing Facebook application platform is also challenging. When Facebook disabled application notifications, they took away a major advantage of the platform. For example, notifications alert readers that someone has commented or liked their story or comment. We had to build our own notification system into our platform, but it’s only effective if readers visit regularly or opt in to receive email notifications. This change also greatly affected how our partners thought about our project, which we discuss in What Partners Say below.
Facebook’s ongoing tone deafness to end user privacy is also a barrier. A number of site visitors hesitate to register for the application because of concerns about privacy. We are regularly asked to allow non-Facebook users to log in to the application.
Over time, we will add other authentication systems to the platform, but in the meantime we rely on Facebook because it does an excellent job authenticating the identity of participants. We do not want our platform to be populated by anonymous readers who are not accountable on a real name basis for their content. This leads to the kind of hostile and uncivil comment threads seen on many major news sites. For now, we believe that the pool of five hundred million Facebook users leave a large audience for the initial experimentation with our platform.
Our interactions with the Facebook Media Team were mixed. Their team is mostly interested in getting news partners to integrate Facebook’s proprietary widgets into their websites. One blogger recently wrote “[Facebook] needs to work more on its actual publisher offering and tools and less on its PR campaign” (http://newscloud.net/edjoj1).
Our platform largely avoids Facebook’s proprietary widgets so that it can be used in the open source community and integrated easily into other systems such as Twitter. So, their team was not very eager to support our work as its open source code base is at cross-purposes with their organizational goals.
The Facebook platform team was also quite slow to respond to bugs. Recently, they began responding to bugs over a year old. And, in general, their first response is to say just about anything doesn’t reproduce rather than make a good faith effort to look into the problem. This is par for the course for large corporate platform providers.
Furthermore, Facebook makes regular changes to the platform that regularly break existing functionality in our application, even the functionality of commonly used open source Ruby plug-ins. They have improved the process of notifying developers before these changes are made but it continues to use up valuable development resources keeping up with their frequent adjustments.
One of the most interesting points partners have made highlights a gap in expectations between us about our project.
As we mentioned earlier, in spring 2010, Facebook removed the ability for applications to send notifications to users. This greatly reduced the viral nature of the default application experience.
Many partners joined our grant project to have a Facebook application provided for them. But, when application notifications went away, they felt that the application didn’t provide them viral advantages.
However, at NewsCloud, we never relied on Facebook’s application notifications to build our platform. Our goal was not to overload the friends of news readers with application alerts, but instead to build new kinds of interactive community features that engage the audience in ways that newspaper don’t today. This is one of the reasons why we’ve often encouraged partners to promote their Facebook Connect websites instead of their Facebook application. See http://newscloud.net/eVfG4e. We even built in pop up and email notifications to substitute for the old application notifications.
But, instead of linking to their NewsCloud-powered community sites, many news organizations just lost interest. For the most part, readers never learned about the new community features.
For me, this highlights how many news organizations think of social media simply as a tool for driving page views rather than as a fundamentally new model for engaging their audience and establishing online community hubs.
One partner spoke of Facebook’s outreach to media organizations which includes proprietary plug-ins:
“The real bad guy is Facebook in my book. When this process began, the concept was for as close to a total integration with Facebook as possible, which newspapers would crave. It was the selling point – a Facebook community news app. But, instead Facebook has essentially made it easy for the media to use plug-ins that get close to a social experience. A plug in that tells me what my friends are reading is ‘good enough’ for most of our audience.”
This is a good example of how our partners don’t seem to envision what interactive features like what a news prediction game or blogging or lending library could do for strengthening community and driving page views over time. Our platform’s set of interactive community features (http://newscloud.net/idLbRa) add so much more than a “what my friends are reading” widget.
Says another partner: “Facebook’s platform evolved away from applications from when your grant was being planned until the point where we were ready to start implementation. In that period, applications went from being what seemed like the core of the Facebook-experience to being used mainly for social gaming. So the experience was unfamiliar to both the audience (who needed to try it and understand it) and stakeholders. At the same time, I am not sure that your straddle between open source and Facebook really works: Facebook users are instantly familiar with how a ‘like’ button works, but you’re not using it. On the other hand, the application doesn’t quite make sense on its own (or at least, was more easily understandable as ‘this is a thing on Facebook’).”
We’ve debated a lot on integrating Facebook’s proprietary Like button, and we may yet still. But, we feel that our current Like button provides more flexibility and doesn’t tie us into a proprietary voting service. We do allow readers to post anything they like to their Facebook wall. We may also allow future partners to choose to make Facebook Likes the default behavior in their application.
Some partners highlighted the resource and mindshare challenge:
“Another factor is that the organizations are risk-averse. Without a proven model of success, there’s little to persuade the organization to mobilize behind something new. Especially to say ‘launch a new community website’. On the other hand, ‘launch something on Facebook’ is more understandable.”
“There is so much going on in the social media space, and so few resources devoted to it. People in my shoes need to decide whether it’s more important to be working on NewsCloud or training folks on Twitter, building Facebook pages, tweeting and Facebooking for the company, building a Tumblr, working on Foursquare, pushing for better promotion of social media on stories, integrating registration with Facebook, pushing for better tweet this and share buttons, trying to measure all these things, trying to check out what other news (and general) sites are trying to do, and trying to see at least a few months down the road. I have a lot of battles to fight for resources and promotion.”
“Last year, we spent about $15-20,000 in staff time or small out of pocket promotion costs on your app. Our gross revenue from Google ads: $156. Now, could we have sold ads on the site? Sure, let’s just say we could have sold $4,000 and gotten some more traffic with heavier promotion. The commission on that, the processing costs, would cut that to about $2,000 net. Hmmmm. Certainly, not everything has to make money, but that’s a tough one. We want to satisfy the ‘Facebook need’ for our audience, but is this the way to do it? The problem is: we need millions of dollars especially in these nutty times”
One partner sums up our work together with a positive note: “From my point of view, I have learned a lot from this and sometimes learning is painful. You have also learned a lot from this, and that isn’t a bad thing. Sometimes the path to innovation and change takes several turns and dips before lightning strikes.”
We were very excited to work with high profile partners such as Carolyn Hax at The Washington Post and The Boston Globe.
The Charlotte Observer found that their community site of bloggers provided an early warning system of sorts for stories which they later reported on their website. See http://newscloud.net/gv4TlA
Minnesota Public Radio’s Paul Tosto described the MinnEcon site as the ideal beat reporting toolkit, relating it to the vision that Jay Rosen has spoken about. See http://newscloud.net/ft2CkD
We added a large list of important community features http://newscloud.net/idLbRa and our platform provides support for key infrastructure features such as international language translation, SEO optimization including site maps and offline notifications via email.
We launched two particularly groundbreaking features: a predictions game, which allows readers to guess about future news events and includes a high scores list for the top predictors and a lending library component to our classifieds solution, which makes it easy for people to share physical goods such as movies, DVDs or household tools with their Facebook friends or the entire community. The lending library includes support for the Amazon product catalog API that streamlines adding items you want to share.
Furthermore, we delivered a wide variety of online support and guides for organizations adopting the NewsCloud platform to read sites. Some of this can be seen at our support site: http://support.newscloud.com and http://newscloud.net/gMvB3A
We rebuilt our code base in Ruby on Rails, which offered many advantages and a cleaner more secure architecture. You can read more about our decision here: http://newscloud.net/huCwIC
In December 2009, we delivered a more easily installable version of our PHP-based Facebook application (http://newscloud.net/hmNspW), which was used for the earlier research grant.
As we look ahead, we’re going to spend less energy and resources trying to manage the level of promotion and adoption by our news partners and more effort into reducing the barriers to entry for open source adoption, mainly making our platform easier to install, setup and manage, improving the design and making hosting more economical. We’re hoping that having a higher number of small adopters will lead to more visibility and adoption of our open source code base. See http://newscloud.net/gisw58.
We’re also encouraging community foundations to work with grantees to provide local online hubs. See the Knight Foundation’s Report Creating Local Online Hubs http://newscloud.net/hmR0Yr and our guide for foundations wishing to have their grantees run social media communities: http://newscloud.net/gMvB3A
We’re also trying to showcase the potential of the platform here in Seattle, with our prototype site, The Needle (See http://theneedle.us). However, we also struggle with limited time and resources for leadership and promotion.
Here are a number of ideas we’ve given the foundation that might make this kind of grant proceed more smoothly in the future:
It’s difficult to provide constructive suggestions to the news organizations we worked with. In some ways, I feel that our report will fall on deaf ears. The very kinds of organizational challenges embedded inside today’s shrinking media companies make it difficult for it to internalize and act on these kinds of suggestions.
From our experience, most simply do not have the capacity to make long-term commitments to this kind of technical partnership. In general, organizations quickly scaled down or back away from the initial commitments they made – even commitments that were carefully screened and documented in our application approval process.
In the future, grantees seeking similar kinds of partnerships should be wary of organizations that jump at the chance to be a part of a Knight Foundation project, but quickly backpedal on their enthusiasm and commitment. We generally saw this across the board.
Our primary takeaway from this project is that most news organizations are not the best partners for this type of technology project. They are just too consumed by a broad set of challenges to their business and unable to engage effectively with a technology like ours.
From our vantage point, the downsizing and budget cutting at most news organizations are dealing a critical blow to their capacity to manage involvement in and invest in new opportunities.
Here are some takeaways we can offer:
Here are some promotional techniques that need to be applied consistently during the initial launch phase of a community site … not using at least some of these techniques guarantees failure (excerpted from What’s required to successfully run our Facebook application community? http://newscloud.net/hxW7Oy):
Here are ten ways to involve readers in the story using NewsCloud’s interactive features:
Ended After Launch
Site Designed, Built, Activated but Not Launched
Note: The Knight Foundation is a financial supporter of the Nieman Journalism Lab.