header image

Analytic advantage online Crunching numbers with captured data
Categorizing user statistics Constructing analysis profiles

One big advantage of new media over old is data reporting. Instead of relying on people's accounts being accurate, data can be captured on the spot. No more relying on Nielson ratings extrapolating data captured from a small fraction of consumers. Old media depended on sampled data, which is more susceptible to problems such as observer-effect or inconsistent findings.
With online viewing applications, exact time can be measured for more accurate ratings. But that is just the beginning. Every user can be enabled to leave feedback for their viewed content, such as allowed on Netflix or Amazon who each utilize star ratings and also allow members to leave personal reviews. But again that is falling into the trap of trusting users.
What about other available metrics? Let's look at extensive approaches a company can take. Many of these would be overly complicated to implement but it's important to understand the breakdowns which would be the most valuable.
The first step is to create a user profile. Depending on the nature of interaction, this can be tied to an account or IP address. Separate from the user account which the user themselves set up within the system, the profile of the user is only used for system analysis. The profile of a user is needed to properly decipher behavior. Different people may exhibit the same specific behavior but their reasons may be entirely different. User profiles can help understand the difference by monitoring behavior over long periods of time.
An example of this would be two people who watch a movie. If they both cancel out after 5 minutes and never resume, it could be assumed that neither enjoyed the content. But what if User A has a history of cancelling a movie shortly after beginning, while User B almost never does this. In this case you'd want the system to assign different weights to each user's actions. By developing user profiles, systems can begin to interpret their specific actions and intentions, and then assign the correct value to the event.
What are some other characteristics a user profile should study and collect information on?
Correlation of Content- Does the user view a wide-range of content? Or can the content often be assigned to a specific category. Depending on the nature of content, there are a plethora of factors to be considered. Continuing with the video example- Length of movie, color palette, pacing, cast, production team, story category, average length of scenes. In an ideal world, a system could collect information for both a user profile and content profile. Content profiles would break down the different characteristics of content material and assign values based on other similar content. By combining user and content profiles, a system can get a better sense of how well specific content would pair with a user, based on their viewing patterns and projected enjoyment of the content.
Viewing behavior- tied into every other aspect would be the behavior exhibited by a user when viewing content. Understanding how users are utilizing services is paramount. Is their full attention grabbed or are they using other services simultaneously. How actively engaged is the user? Combined with length of use and occurrence, user satisfaction can be measured. User A uses the service every night for 3 hours. User B uses the service once a week for 2 hours. Which has the higher satisfaction? Easy answer is User A, but a detailed user profile with viewing behavior can tell you that User A uses the service in the background and out of disinterest, while User B is more actively engaged in their experience. These complicated answers can be found through detailed data of viewing behavior.
User peer similarities- The biggest benefit of user profiles is combining the different profiles, building a clearer picture with aggregated data. With individual user profiles, you can begin to group together demographics. Netflix can understand which titles have wide appeal, which have cross appeal, or which are only specific to small groups. Understanding this difference makes the decision of which projects to fund much easier. The idea is to break users into as many independent groups as possible and then identify the content with the greatest reach. Individual user profiles let you make dynamic groups based on factor. User A and User B both always watch serious movies under the category drama, and both watch roughly 5 titles a week. Sounds like they're in the same group. But when funding projects you'll want to also look out for possible benefits from indirectly related groups. Maybe you're producing a dark comedy which centers around death, and exhumes the same slower pacing found in dramas. At that point only User A or User B might be a potential viewer. Using links between users with different preferences is difficult but perhaps the most valuable metric. That's the point where you are finding valuable content the user themselves are not yet aware of what they'd like. In other words, recommendations that can be trusted.
Instead of relying on people to interpret their own motives or answer accurately, studying patterns in user activity offers advanced benefits. Record user interaction with the system, identify areas which show the best results and consumer value, funnel funds towards those areas and accumulate data the entire time. That is the dream which new media has realized. Of all the benefits New Media has over old, this one receives the least attention, but carries the most weight.