The Video Audience Measurement Framework

We measure video audiences primarily to:

  • Understand preference for different types of content
  • Measure exposure to any advertising that has been placed adjacent to or embedded within that content so that it can be priced fairly

As the modes of access to video content  (in terms of both, distribution and devices) and advertising increase, the measurement of audiences and advertising exposures become increasingly challenging.

Let’s start with linear and time-shifted television. These were probably the easiest to get off the ground. We started by asking people to recall which channels they watched recently/ past week and yesterday or asked them to maintain a diary to record the channels they watched and at what time. As the number of channels proliferated, a diary or a recall-based method became unviable. To overcome this challenge, meters were introduced. These meters or “people meters” measure and report two pieces of information – who at home are watching and what was being watched. There are essentially two types of meters that are available – Portable meters and Fixed meters.

Fixed meters are either connected or placed close to the Television set at home, based on the underlying technology to identify the channel/ station being watched. Portable meters are carried by panelists all day and the device automatically records any exposure to Television content using ACR – automatic content recognition technology. There are two alternatives here – watermarking – a unique code that is inserted into the content at source – inaudible to human ears, that enables us to identify the content being watched and fingerprinting – where unique fingerprints are created for the content using pre-defined algorithms and these are matched with a reference library of fingerprints to identify the content being watched. While watermarking works well with live television content, fingerprinting is more suitable for digital video content which is consumed across multiple devices and asynchronously.  Portable meters have an added advantage in that they allow for the measurement of video content consumed out of home as well.

As broadcast technology has evolved, the digital set-top box is capable of storing and reporting the TV channels the box has been tuned to and even other activities on the remote, such as the press of any interactive button or change in volume. Telecom or internet connectivity allows for this data to be sent to a central location for near real-time reporting. The advantage of such an approach is that it is passive. The disadvantage of such an approach is that there is some uncertainty about the TV set being switched on. If the box is on and tuned to a channel but the TV set is off, it does not constitute viewing, and who at home is sitting in front of the TV set, if anyone is at all

The advantages of using a set-top box return path are that no capital expenditure is incurred in building any meters and installing them in panel homes, allowing for a sample size as large as possible can be processed on a daily basis (the census approach). However, the two uncertainties identified above need to be addressed. The first one can be addressed using technology – an HDMI connector can be used to identify if the display (TV Set) is on. The second uncertainty can be addressed by using a diary/ meter-based panel (run in parallel with smaller sample size) for calibration and attribution to age, gender, and other socio-demographics.

Another change in the technology with the evolution of set-top boxes is that they can serve different advertising to different households. Once such boxes become prevalent, panel-based audience measurement will be completely inadequate to measure advertising exposure – which is one of the key requirements it serves. Return path data also allows for measurement of advertising exposure in an addressable TV environment where different subscriber homes are exposed to different advertising – based on location and household profile.

Given the large number of households that can be covered in any RPD based audience measurement, there are a few considerations to be made:

  • Data privacy – Do households need to formally opt-in to share their viewing behavior or can this data be collected just as websites and social media platforms collect data about their users? How do we ensure that when such data is used for reporting audiences, it is anonymized
  • Understanding of the household profile – some understanding of the household size and profile of the household members is essential to be able to assign viewership to individuals in a household
  • Participation of the Cable/ DTH operators who can collect and control access to this data. There is one critical element here – the data from the STB must be encrypted and reach the independent research agency directly to ensure that it cannot be tampered with, I anyway.
  • While RPD means an opportunity to gather viewership data at a census level, in some situations, a sample-based RPD may make more sense. For instance, factors like the technical capability of the STB to return data, the cost of processing large volumes of data, and the ability to be able to attribute household-level viewing to individuals at home.

Sampling

Prior to proposing a sample size and design, it is important to understand the TV/ Video viewing universe. An establishment survey must be conducted for this purpose at least once a year, as the TV/ Video universe is changing quite rapidly with a range of devices and multiple modes of access.

Sampling must ensure representation by

  • Mode of signal reception: Terrestrial/ Cable/ DTH/ IPTV
  • Operator
  • Internet connection at home
  • Subscription to multiple modes of receiving content, OTT
  • Geography
  • Language spoken
  • Family Size
  • No. of TV Sets at home

It is important to ensure that the representativeness of the sample is not just at an overall level but at each reporting unit.

Technology

Audience measurement is increasingly driven by technology – identification of channel or content or real-time collection of data from meters/ STBs must be done efficiently, securely and free of error.

Smart TVs

Smart TVs are also equipped with ACR technology and can collect fingerprints for the identification of content consumed. This data resides with TV manufacturers such as LG, Samsung, Sony, Vizio (Inscape). For example, the smart TV penetration in India is at about 20 Million and this is expected to grow rapidly over the next 2 years.

Cross-device Audience Measurement

Video consumption on other devices such as computers and mobile devices can be measured using watermarking and fingerprinting technology. Alternatively, digital tags can be incorporated into the content and detected by software-based meters installed on devices on which the content is consumed.

As budgets for audience measurement are bound to have limits, it is important to have the right priorities set: For instance, is it more important to measure viewership within each medium more accurately or is it more important to measure the duplication across media more accurately?

Measuring more than exposure

Audience measurement currencies are designed to measure and report exposure to advertising and content. These are set up to be representative of the market – say a country or a state. Increasingly, there are alternative measurement services/ solutions that are available that

  • Measure ROI – return on advertising/ media investment
  • Research solutions that measure the shift in brand awareness/ intent to purchase/ shift in endorsements of specific brand attributes, sales – this is done not just on Television but across media
  • Optimize cross-media campaigns by using TV exposure information to deliver advertising on mobile devices to those who have not been exposed to advertising on TV

The Roles of the Players in Audience Measurement Ecosystem

Traditionally, media research companies have been data creators for the audience measurement ecosystem, and broadcasters, media agencies, advertisers and cable/ DTH operators have been consumers. That has now changed as broadcasters, operators and ad-tech companies have become data creators as well. Furthermore, such data is more granular and in larger volumes – actual rather than an estimate.

Collecting anonymous STB return path data can be valuable to cable/ DTH operators for understanding the content preferences of their own subscribers and this data can be monetized by supplying to audience measurement companies.

Thus, the audience measurement system is no longer limited to a media research company, which will have to work with a number of other companies – online platforms, cable and DTH service providers who have the ability to provide census level information and a comprehensive view of how video content is being consumed.

The Human Element

The audience measurement activity needs to be undertaken with minimal intrusion i.e. it needs to be as passive as it can be so that the act of measurement itself does not alter the viewing or its reporting.

  • Minimal visits to/ calls/ contact with a panel member
  • Minimal action – logging in as a viewer on a meter/ bringing up or opening the app on a mobile device/ changing any settings on the measurement device/ mobile phone

The audience measurement agency must provide assurance on the identity of the household and personal data privacy.

Any meter-based audience measurement also needs to ensure full compliance on the part of the panelists – be it registering viewership on the fixed meter using a remote or carrying a portable meter with him/ her out of the home.

The risk of a broadcaster/ content creator attempting to influence a household in watching a channel/ program is always there. The audience measurement service provider must always watch out for changes in household viewing behavior that cannot be explained.

Granularity of reporting

Depending on the sample size, it is important to allow for as granular reporting as possible. In an event a compromise must be made, meaningful grouping must be created based on state boundaries and or linguistic majority.

Time-shifted viewing within a reasonable frame, say 3 or 7 days, must be incorporated into the currency reporting.

Any panel-based reporting must also provide the corresponding standard error associated with it.

Conclusion

Any audience measurement system – passive or active, must be trusted by all users. Building trust is important through

  • Participation of all stakeholders in any critical decision making – research design, sampling, software, processing rules- defining outliers. Decision-making should not allow for bias of any sort.
  • Third-party audits by industry experts.

Any audience measurement company must ensure that it is transparent with algorithms used to identify outliers, the rationale for exclusion of any household in reporting, and formulae used in the computation of any metric – reach, rating, and time spent.

It is likely that broadcasters, advertisers, and agencies will move away from using a single currency that will meet all their requirements to using multiple services that will be required to meet their needs. Currencies are important and will remain relevant, but they would need to transition from measuring just Linear and Time-shifted TV to include BVOD, OTT, and Online Video platforms as technology enable easy access and audiences adopt them. The measurement also needs to expand the devices used in addition to Television – Mobile phones, Tablets, and Computers. It is important that this transition of measurement should be initiated now so that the measurement technology can keep pace with media consumption patterns.