13 Principles for How Media Companies Should Use Data

    by Dorian Benkoil
    March 28, 2011
    A recent survey by media conference company DigiDay asked people in the advertising industry if "data leakage" was a key concern.


    It's easy to get excited over a spike or drop in traffic, visits, streams, sales, downloads, or registrations. But that spike, over a day or a week, often doesn't tell the trend.

    Business content on MediaShift is sponsored by the weekend MA in Public Communication at American University. Designed for working professionals, the program is suited to career changers and public relations or social marketing professionals seeking career advancement. Learn more here.


    Access to, and control and ownership of, data is playing a big role in everything from debates over privacy to the new Apple subscription plan to publishers’ negotiations with advertising networks. There’s a very good reason.

    For a media company, data is lifeblood. If a company gets the data and gets it right, it can increase audience and revenues, generate ideas, and continually improve and compete.

    Media executives are increasingly realizing the value of data in the digital realm. AOL paid a premium to acquire the Huffington Post in part because of HuffPo’s skill at gleaning and acting upon its user stats.


    “The biggest negotiating point in contracts these days” is data, said an attorney for an online advertising management company at recent a privacy breakfast hosted by Content Next.

    So, if you’re a digital content company, how can you best make use of your data? To help you maximize your use of data, here are some key principles and cautionary notes:

    1. Get It From Multiple Sources

    It’s surprising to me how often someone asks for my company’s help with digital media but can’t answer even basic questions about where they currently stand.

    Make sure you have not only a web analytics program (Google Analytics and Omniture are two of the most popular) to measure traffic internally, but also other measurement systems such such as from Quantcast, Compete, Nielsen, and comScore. These tools can give you additional information on the demographic composition of your site visitors and how your site traffic compares to others.

    The more types of data you have, the more full picture you can get. If you serve ads, make sure you know your ad serving data, too. (More on that below.) Try to work with providers who share data with you that they collect on your users.

    Apple has come under fire from some publishers for limiting access to data related to subscriptions sold through its App Store. This is a very real problem for content companies.

    “Access to the audience is crucial,” said Bob Carrigan, CEO of IDG, the largest tech trade publisher. At the recent Bloomberg Media Summit in New York, he spoke about how the lack of data available from Apple’s announced plan was a problem.

    “The issue is building a direct relationship with our users so we can give them what they want,” he said.

    Pam Horan, president of the Online Publishers Association, agreed. “Apple is going to potentially limit the growth of the market” by limiting publishers’ access to data, she said.

    2. Know Your Goals and Measure Progress Toward Them

    The wrong data can be worse than useless. It can misdirect your efforts.

    Having a lot of pageviews, for example, is certainly nice. But are they the right pageviews? What if your aim is to get as many people as possible in your community to look at your website, but your visitors are coming from outside the community? What if you make most of your money from ads placed inside videos that don’t require a new pageview to watch? You’ll want to know not how many new pages people look at, but rather how many videos people are watching, and for how long.

    Donald Chestnut of marketing agency Sapient Nitro noted at the Media Summit that sometimes the data a publisher has is not what he needs. “Publishers have a lot of data, but it tends to be demographic,” he said. “As a designer, I need behavioral information” on how people are using a website, internally.

    3. Cross Check

    A big reason you want different types of data (as in point 1 above) is so you can use different measures to figure out what one data set might not be telling you.

    On MediaShift, for example, I discovered that the pageview growth we saw in Google Analytics wasn’t matched in our ad serving statistics. We investigated and discovered that some key pages visited by large parts of the MediaShift community didn’t contain the code to serve ads. We have since corrected the error.

    4. Look at Your Data Comprehensively

    It’s easy to get excited over a spike or drop in traffic, visits, streams, sales, downloads, or registrations. But that spike, over a day or a week, often doesn’t tell the trend.

    Data trends are more important to longer term success of your site, your app or anything else.

    I use moving averages, logarithmic graphs, and other “smoothing” techniques to recognize trends over time. I’ll also sometimes eliminate an outlier, such as an unexplained daily spike up or down for one story, to get a cleaner picture overall.

    5. Look at the Calendar (and the Weather)

    This could also be called “why humans matter.” Often, it’s logic and thinking that will solve a quandary and help you grasp what’s going on.


    For example, at a suite of websites I helped manage, I was having trouble figuring out a dip in traffic compared to the same period the previous year. Then a smart colleague suggested we look at the calendar.

    It turned out that for the months I was looking at there had been more weekend days than during the same dates the previous year. Since people tended to access the sites from work, we seemed to have the anomaly explained. I re-did the calculations to account for the extra weekend days and the drop was accounted for.

    I’ve seen other traffic swings caused by things like an election or harsh weather. Think of factors having nothing to do with what you’ve done that may cause traffic fluctuations and then try to account for them in your thinking.

    6. Accept Uncertainty

    MediaShift executive editor Mark Glaser and I have repeatedly gone over a spike in traffic one day last month that we can’t account for. No one story or search term or website gave us a flood of traffic that day.

    There was no big news or mention of MediaShift or one of its stories, as far as we know. Was it a glitch in our analytics software?

    We may never know. That may be one of those outlier days I smooth out when looking at the longer-term trends.

    7. Understand the Trends

    Again, this is a “humans know best” point. I sometimes prepare charts that show not only traffic spikes and dips, but also correlations for those dips.

    Then, I examine a series of questions to explain them. For example, does this site have a seasonal trend, such as a dip or flood of traffic during the year-end holidays? Are there promotional campaigns running? Is there something in the news that got people searching on a term? Have more stories been produced, more feeds put out?

    All of which leads to a corollary:

    8. Map Your Data to Other Key Factors

    To do this, you must also (as I mentioned in the second point) know your goals.

    Some questions to help assess your goals: If a traffic increase has come about because you’re providing more stories, how much is it costing you to produce those extra stories? Is the extra traffic you see coming worth the extra resources you’re using over the longer term? Is the extra traffic the kind of traffic you want?

    9. Test and Retest

    One of the beauties of digital media is that it can be almost cost-free to try things. Even small improvements, continued over time, can lead to big changes.

    Japanese manufacturers use the term “kaizen,” or continual improvement, for this practice. It’s how, through myriad small tweaks, Nissan changed its cars over decades from shaky boxes to the pinnacle of luxury with its Infiniti line.

    Be sure, though, that before you change things, you measure where you are, see what the data is telling you, and then collect data again for an appropriate period of time after the changes to see how they worked. Try also to isolate the different components and factor in other things (such as anomalies) that may have contributed.

    10. Should You Share?

    i-d3723cc7d8b8e8fa39b0430d9f8141cf-Screen shot 2011-03-27 at 8.49.50 PM.png

    Gawker Media from its early days has chosen to share data about pageviews and other factors, as you can see by searching the site on Quantcast.com. WashingtonPost.com, by contrast, has chosen to hide similar data from public view.

    This one is a big question, for which it’s hard to give an absolute answer. The spirit of openness says you should reveal your data, rather than create suspicion that you’re hiding something. On the other hand, there may be something about your site that you’re afraid the general public — or at least your target business markets — may misinterpret. So, instead, you’d rather have them ask you, and have your staff explain the stats.

    I’m not doctrinaire on this one and have advised sites both ways, depending on their individual strategies.

    While data is a helpful tool, it’s not the end-all and be-all, nor should it replace creativity, logic or thinking. Here, then, are a few cautions on the limits of what data can do:

    11. Don’t Let Data Stamp Out Creativity

    Data is a “left-brain” way of thinking — math and logical processes.

    Creativity often isn’t. The data-driven way of spurring creativity is certainly valid. It’s why we have so many sequels to hit movies from Hollywood.

    “Hey, that film did well,” studio executives probably think to themselves, “so we need to do another one!”

    You get lots of Spiderman and Batman movies, and some are actually pretty good. But that data won’t spur executives to fund “Brokeback Mountain” or “Syriana.”

    We might some day have a predictive model that can figure out what it was that led director Darren Aronofsky to successfully use Mickey Rourke as the lead in “The Wrestler.” But we certainly don’t have that today.

    Nor do we have a way of knowing for sure what out-of-the-box story or video will be a surprising hit.

    12. Beware of the Short Term

    People in the media business, especially the news business, often obsess over immediate results.

    They’re obsessed by questions such as a TV show’s overnight ratings, how many pageviews a homepage headline is generating right now or what a film grossed over the weekend.

    By those kinds of short-term measures, “Star Trek” and “Seinfeld” were only modest successes in their earlier days.

    I’ve frequently seen feature stories, slideshows and other media on websites that don’t appear especially successful at first but over time rack up significant traffic and turn out to be among the site leaders.

    Make sure to pay attention to the longer term in measuring what’s working.

    13. Do Things That are Worthy, Regardless

    Sometimes a certain story, video, song, etc., is the right thing to do just because it is — perhaps it is emotionally rewarding to you or staff, or reveals some important information for a part of your community.

    Sometimes you just want to take a chance and do something no data, no focus group, no logic would tell you should be done. (Like to make a list of 13 — lucky! — bullet points.)

    Don’t believe that any data is the only way to measure success.

    A former award-winning managing editor at ABCNews.com and an MBA (with honors), Dorian Benkoil handles marketing and sales strategies for MediaShift. He is SVP at Teeming Media, a strategic media consultancy focused on attracting, engaging, and activating communities through digital media. He tweets at @dbenk.


    Business content on MediaShift is sponsored by the weekend MA in Public Communication at American University. Designed for working professionals, the program is suited to career changers and public relations or social marketing professionals seeking career advancement. Learn more here.

    Tagged: analytics comscore data gawker media google analytics omniture pageviews quantcast

    Comments are closed.

  • Who We Are

    MediaShift is the premier destination for insight and analysis at the intersection of media and technology. The MediaShift network includes MediaShift, EducationShift, MetricShift and Idea Lab, as well as workshops and weekend hackathons, email newsletters, a weekly podcast and a series of DigitalEd online trainings.

    About MediaShift »
    Contact us »
    Sponsor MediaShift »
    MediaShift Newsletters »

    Follow us on Social Media