What's Your Video Data Roadmap?
At this year's CES, a Sony executive talked about how the company tested four different types of native ad movie trailers within Facebook for different markets. One did so well that they decided to run it as a television commercial.
None of this would be possible with without the insight from detailed analytics. Analytics can answer myriad questions: Are publishers running too many ads with their content? Would viewers watch more if the content was shorter? Is ad inventory priced correctly? Effectively using the data from each part of the technology stack means better business decisions and happier customers. This article introduces a quick overview of the data roadmap that can direct publishers to making the right decisions about content and advertising.
For example, the staff at eight-year-old news and current events site Newsy built its own analytics dashboards. "We have dashboards for revenue, we have dashboards for editorial teams. I look at it multiple times a day," says general manager Blake Sabatinelli. Newsy also has a large syndication business and deliver ads programmatically via first- and third-party data.
"We build a lot of our stuff in Excel and Google Sheets," says Sabatinelli. Newsy uses Google Analytics and Supermetrics for acquisition and automation. "We capture all that data that comes out of our CMS—length, tone and tenor of story, sentiment (happy, sad, or neutral story), the number of pieces, who are the top authors, and merge it with data on the other end, which is how our customers are accessing (content). We try to do everything we can to merge the two data sources to get insight into how everything is working."
The Beginning
For any publisher, the journey begins when a user starts a video. "The three most common quality of experience (QoE) measurements are how long did it take for the video to start when the consumer clicked play, what bitrate is the picture, and what was the rebuffering ratio—how much time did they see the buffering wheel vs. how much time did they watch content," says Ed Haslam, chief marketing officer at analytics service provider Conviva. These measurements may not be news, but having all the pieces in place however is news to many organizations, so leaving this out would be like starting from step two.
Being able to track content wherever it's published is the next step. "One of the thing popular with our customers is social syndication. The days of 'I'm going to put my video out and it's going to live on my app or website or Roku channel' has evolved to include social media," says Matt Smith, vice-president and principal media evangelist at online video platform Brightcove.
Percent Complete
The total percentage of the video viewed and where it was viewed are vital stats almost all publishers track. If there's a big drop off at a certain point, there will be platform data on connectivity issues, but the problem might be something simpler—maybe the content wasn't interesting.
On the social side, some companies customize content for each platform and others don't. Either way, publishers need to know if people are engaging with it, sharing, liking or retweeting it. "Being able to pull the data back into the platform and put it into a dashboard so our customers can see that one piece did really good on Facebook, OK on Twitter, and not so good on Snapchat is very important," says Smith.
Data provides grounds for hypotheses. Whether a given hypothesis is right or wrong takes human analysis, but without the data, publishers are only guessing. Newsy found that the idea that viewers wanted to specifically curate their own content was more popular in theory than in practice. "In publishing I think there is this fallacy that a large amount of people want to pick and choose. At the end of the day, that audience is usually very narrow," says Sabatinelli. "Instead of forcing people to log in and pick and choose what categories you want to see, the simple act of paying attention to what people are consuming via data and doing passive customization around the content and the ads is a more effective experience."
Ad Break(s)
Switching from the viewer to the advertiser, video marketing platform Innovid has analytics that drive hyper-personalized ad campaigns and provide insight about who was targeted, how many ads were delivered, to what devices, the completion rate, when and where it was viewed, which area of the site it was viewed on, whether it was viewable or not, and what steps were taken when interaction occurred.
"If you're Bank of America, you have a lot of types of customers and there's a lot of ways to tell a story," says Tal Chalozin, chief technology officer at Innovid. "Because of all these permutations, the connections of the data must be automated because there's no way to do it in a manual way. We are connecting video (data) into the marketing automation platforms. We call it data-in and data-out. Data-out queries the marketing cloud a publisher may use to find out details about the viewer. Data-in delivers whatever information around your identity is acquired when you view a piece of content back into the marketing cloud."
Let's say a piece of content achieved 10 million impressions across five days, twenty sites, and five different devices. Nielsen has data about the GRP rating—the reach, frequency, the audience composition of demographic and psychographic data. "(We) clean it and connect it into all of our other data sets and hash it all together." says Chalozin. Innovid works with third-party data from ComScore, Nielsen, Moat, Integral Ad Science, DoubleVerify, and others. The endpoint is knowing how many impressions were delivered to the right audience.
A number of companies, including Innovid and Furious Corp., help customers normalize their data. Essentially they're offering a cleaning service to take first- and third-party data (Innovid) or data from disparate systems (Furious Corp.) and help process it into a more standardize format which can be imported and used in other systems.
Moving from data cleaning to ad volume brings us back to Conviva, which provides measurements to gauge if the volume or placement of advertising may be mismatched with consumer expectations. In other words, it helps answer the question "How many ads are too many?" "The metrics we're providing help them decide on the best placement for ad pods and how long should an ad pod run. They can vary those and measure overall engagement. Publishers can take this knowledge and do A/B testing to see what resonates with viewers," says Haslam.
Data Access
All the vendors interviewed here have their own dashboards or hooks into the data where customers can export information. "You can connect through an API and get a real-time feed of the data, or do a batch transfer where you save whatever time period you're interested and transfer via a file, which could be an Excel-like file or some other XML-based markup transfer file," says Haslam.
Publisher technical operations groups can connect the data feed to a monitoring tool like Grafana to ensure there is uninterrupted video signals. Or product or marketing groups may prefer their data to be routed to a visualization tool like Tableau or Splunk or to a marketing cloud where the information can be combined with a CRM database. This intelligence gained from viewing trends can provide guidance on what features or services to invest in or how to market more content.
A current trend is the emergence of hybrid delivery. where some content is AVOD and some SVOD. "Crunchyroll is a good example of that where you have an entry-level product that is ad-supported and then you're able to upgrade to a subscription-based product," says Haslam. "We've seen multiple publishers start to play with both ad- and subscription-based models to try to suss out the consumer's appetite for doing both or doing more of one or the other." Having data means being able to see what consumers are partial to.
On the one hand there are companies at the start of the data roadmap. "My advice is for people is to get religion around data. I think you would be surprised at how many media organizations don't really get deep in the woods with the data. They may look at it and say, 'Yeah our views are up Tuesday or Wednesday', but they won't look into individual pieces of content abandonment rates, (and ask) 'Did we play more ads on a particular stream?'," says Smith. "Use the data to understand why your viewers are engaging (and) how they're engaging, and use that data to drive a video strategy."
On the other hand, there are broadcasters that are consulting their analytics several times a day to see what is resonating. For Newsy, getting real-time analytics is the big ask. "The one thing I wish we had is true real-time data coming off of live streams," says Sabatinelli. "Nielsen has to wait for overnights to come out, to understand you had a drop off at fifteen minutes after the hour. Having that real-time data across all platforms would be the most powerful tool imaginable." For Google Analytics users, Charles Farina, manager of digital analytics at Analytics Pros has a solution that can be integrated into the player that he says provides data on concurrent livestream users.
The Journey Ahead
Furious Corp. works with publishers to bring automatic consolidation of data sources from disparate advertising inventory management, planning and pricing solutions, replacing the manual approach their customers were previously using to run their ad businesses. When talking to technology vendors, says Furious Corp. CEO Ashley J. Swartz, the first question you ask should be "'Do I own my own data?' The follow up then is 'If I own my own data, how do I get it out?' Ideally all vendors have API access." However Swartz says that's seldom the case in her part of the video tech stack. "At a minimum, I want to get (data) out daily. The third question should be, 'How easily can I access this data and in what format?'"
The number of data points needed vary by publisher and business focus. Did the video play well, where did it play, and who watched? How much of and where was the content viewed, where did viewers drop off, and how many ads are too many? Can viewer information be brought into other analytics or marketing platforms? Is the data current? Is it accessible by API?
On the video roadmap, data is a journey, not a destination. Figuring out the right data to track means extracting business analytics will become a lot easier and that should be good news for both publishers and viewers.