Opportunities for Analytics in Media

Analytics skills are highly relevant in the advertising and media sector, finding a number of applications.

For Publishers (broadcasters, OTT platforms), analytics plays an important role in

  • driving viewership (consumption of content)
  • increasing the subscriber base and minimizing churn and
  • monetizing advertising inventory through appropriate pricing and maximizing the inventory utilization.
  • Forecasting viewership or estimating demand for a particular content and assigning value to advertising and sponsorship for total monetization of content over time

Advertisers are concerned with optimizing their advertising spends and minimizing wastage. Any advertising that does not lead either to building awareness for a brand, intent to purchase it, higher affinity for the brand or a sale, is a waste. Buying advertising needs a deep understanding of the viewer profile, the context in which advertising exposure occurs and the ability to optimize the media plan.  Advertising exposure data and analytics needs to available quickly (in near real time) and the total no. of exposures for a campaign (frequency) to a viewer must be managed, to minimize wastage and fatigue. This is becoming complex with real time bidding using automated platforms for video advertising inventory and newer delivery mechanisms such as addressable advertising through connected boxes and TVs.

Content creators need to assess all the platforms they can distribute their content to, such that they are able to get the best return on their investment. They also need to have a deep understanding of audience preferences, performance of various genres of content, popularity of actors.

None of these goals can be achieved by the stakeholders without the participation of audience measurement companies, monitoring agencies, analytics and data visualization service providers. With such diverse applications, analytics opportunities emerge on all sides of the spectrum with Broadcasters, Content creators, Distribution platforms, Websites, Apps, Data Management Platforms, Media Agencies and Advertisers.

In this piece, I have illustrated some of the potential applications for analytics in the Television & Video space.

Analytics in Media

Media Consumption (Audience Measurement)

Audience measurement starts with measuring access. In some cases, actual data is available (think apps and set top boxes) but these could also be estimates that are derived based on representative sample surveys projected to the universe using larger data sets such as census. For example,

  • Number of households with TV sets and an ability to access the content that is free to air and those that are connected to a Cable, IPTV or DTH set top box
  • Number of individuals with connected devices: computers, tablets, mobile phones and smart TVs
  • Number of downloads and active users for an OTT app (including user generated video sites)

Once access is established, we can focus on analyzing consumption.

  • How many people watched (or listened) to the content
  • What types of content (genres) do people tend to watch e.g., drama, sports, news, movies
  • Events with live telecasts – large sporting events such as the Olympics, Cricket and Football and extensive election coverages boost advertising value and volume significantly, creating more opportunities for analytics
  • As viewership gets fragmented, understanding aspects such as engagement, attention, profiling (household members who watch a particular program together) become important.

Analysing Viewership Data

Any audio/ video measurement needs technology (short of asking people to note down what they watched). The STBs (Set Top Boxes) can capture and report at scale, what is being watched by a household. The fields that are evident in such data are:

  • Session Start Time & End Time (this is in seconds and in a specified time zone)
  • Code for the channel the box is tuned to for that time-period
  • The audio language the channel is set to

The tuning data collected from such STBs is by no means perfect and requires data science expertise to correct for tuning that is unlikely to be viewing, outlier behavior and viewer demographics need to be predicted before it can be used either independently or fused with other data such as that collected from meter based panels.

Online Content consumption via OTT platforms e.g. Amazon/ Netflix/ YouTube can also be measured through logs that are available at the publisher’s end. The data format is typically XML or any text delimited format and the fields that are evident in such data are:

  • Device details (Operating System, Model)
  • Identifiers for the Device, for example: AdID (Android), IDFA (iOS), STB (e.g. Roku ID in a non-web environment e.g. Roku Devices)
  • App or Browser details
  • Play Start & End time (in seconds)
  • Content ID – Whether the content is a live show, VoD or an advertisement, this can be identified by labels
  • Campaign or visitor source identifier

Based on these data, certain metrics can be derived. For example:

  • Total no. of visits to a page that hosts the content.
  • The total time spent by each user on the page.
  • Plays – Attempts to playback content that result in a successful playback. (named as play view) — it is calculated as content start and stop.
  • Playlist completion or Playlist completion ratio indicates the extent of viewing for a particular piece of content.
  • Play Duration – Sum of time of actual playback across all plays. This excludes ad play duration, forwards, rewinds, repeat playback and pauses.
  • Video Concurrency gives the real time count of the no. of video playbacks being made for a particular video title at any given instant.

Engagement

While a publisher focuses on ensuring engagement with the content and generating advertising revenue, advertisers seek assurance that viewers have actively engaged with the advertising. The ability to analyze minute by minute viewership will help answer a number of questions. For example,

  • For TV channels, music streaming sites and OTT platforms that are predominantly dependent on subscription revenue, retaining subscribers is essential. It is therefore important to determine viewer/ listener engagement with content and the app (as these can help predict and manage customer churn)
  • How was the viewing behavior – was the viewing continuous or with a lot of pauses or switches? Did the viewers pay attention to the content or were they distracted?
  • Personalization based on content preferences increases customer engagement with a service or app. STBs are also being enabled to let a user search and look up content of interest across all platforms. They are also able to apply analytics real time and run recommendation engines based on viewer behavior. The machine learning model for recommendation engine is continuously improved through A/B testing.

Advertising exposure

Advertising exposure is a part of all audience measurement datasets such as Linear TV, Ad supported OTT, Online streaming and Mobile. Analytics using advertising data is essential as marketers spend billions of dollars on it.

  • Understanding the optimal frequency of exposure to a particular campaign and the corresponding return on investment from different media (to sales and shifts in brand measures) is especially important.
  • Sometimes advertising exposure is not obvious as brands are often integrated into the content. In an online video environment, such brand integrations can be delivered dynamically. AI allows us to do both – deliver such advertising in a contextual environment as well as measure any exposure to it
  • Addressable advertising allows ad serving companies to deliver highly targeted advertising (based on demographics, affluence, interests, context, and location) to viewers. Thus, different households watching the same show on the same channel can be served different ads. The role of analytics here is to be able to determine these targeting parameters
  • Data science expertise is also needed to combine disparate data sets (Linear TV + Addressable TV + OTT + Online Video) providing information on advertising exposure, correct for duplication and accurately report (or estimate) the overall frequency of exposure to advertising from multiple sources.

The analytics capability – to process, visualize and interpret all such audience data can help publishers drive subscription and/or advertising revenues, advertisers to optimize their investments and content creators to maximize monetization of their content.