HKTAM, a TV audience measurement service, involves estimating the number of viewers tuned into a TV content at any given time as well as their demographics. To obtain the number of viewers and their demographics, panels are deployed. Panels, which are at the core of any TV audience measurement, are used worldwide.
Nowadays, TV viewing is not only restricted to TV sets. It can be accessed via PCs, laptops, smartphones, tablets, and set-top boxes, etc., which leads to the increased opportunities to source data directly from connected devices, aka “big data”. Although big data is considered by many as more genuine and reliable, panels still have a role to play in audience measurement research.
It is undeniable that big data has a lot of merits relating to economic and scientific efficacy. However, when applying it to TV audience measurement, there are challenges and limitations. Some of these are highlighted below:

•    “Data silos” – Despite the proliferation of viewer-related data generated directly by streaming platforms and set-top boxes, this data may be managed by different departments within an organization, making it difficult to access, share and integrate different data sets within the same organization. This data is also broadcaster specific and not for public release.
•    Data quality – Before big data can be used for analytics purposes, we need to ensure that the data is accurate, appropriate, and in the proper format for analysis. If this is not being done, the insights derived by these data can be useless or even harmful. For example, for streaming data, they are often subject to the problem of invalid traffic (e.g., bot traffic, malicious traffic). Another example is return-path data. Kantar has been working on for over 10 years to refine the algorithms needed to clean the set-top box and server data, suppressing those occasions when the box has been left on and no one is watching.  
•    Missing key data – While broadcasters can obtain from their own server logs the viewing volumes of different TV contents, the server data alone misses key pieces of information:
-    Demographic profile of viewers. Ultimately, it is the people not devices or households that media owners and advertisers are interested in. Though demographic data might be available in some of the data sets, very often, they are limited to simple age and gender demos.
-    Cross platform and device deduplicated reach. E.g., How many smartphone users also watch the same channel or advertisement on a TV set, and how many on smartphones only.
•    Incomplete data – Big data are data based on existing customers. However, very often, media owners would need to understand the market from a larger population – potential as well as non-customers. Then it becomes essential to expand the base to cover these groups. In addition, for non-internet-connected TVs, there is no “big data” as such.
•    Data impartiality – Big data, such as streaming traffic data or set-top box data, often comes from the media owners themselves. These data would need to be validated by a third party before they can become a trusted source.

Nevertheless, audience measurement should move at the same pace as the audience and the rate of technology. Currently, there are a number of markets running cross-platform audience measurement, utilizing both big data and panels, by CSM, Kantar and other research providers. For cross-platform measurement, not only a representative panel is at the heart of core TV set measurement, panels also play an important role in measuring internet-connected devices and set-top boxes – panels deliver a vital and trusted source of demographics to bring big data to life, provide de-duplicated reach across platforms and devices. By leveraging a people-based source of audience demographics, advertisers can also enrich their targeting and addressable tools to ensure the optimum balance in media plans and enable in-flight campaign optimization.
In a nutshell, both big data and panels have their merits and shortcomings. Harnessing the power of big data and the accuracy of panels together is something that CSM and Kantar are doing in other markets to report TV-content viewing beyond TV sets.

Sampling is also widely used in academia and in government. In Hong Kong, the Census and Statistics Department conducts many surveys using samples, such as the General Household Surveys and Thematic Household Surveys. These surveys provide valuable information to the government for reviewing and formulating various social and economic policies, and to the private sector for business and research purposes.

There is an internationally accepted standard in audience measurement – Towards Global Guidelines for Television Audience Measurement (GGTAM) produced by the international joint industry Audience Research Methods (ARM) Group. GGTAM is an important guide to best research practice, and provides operational level advice on how to design TV audience measurement surveys, and to promote good practices and principles for audience measurement.
In addition, there are international organizations that provide guidelines in research practices and definition of terms, such as Media Rating Council (MRC) and Interactive Advertising Bureau (IAB).
Though the needs in audience measurement may vary market from market, CSM and Kantar take into account the core principles set out in GGTAM, and make references to other international organizations when designing and executing the HKTAM Service.

Decision on panel size is the process of choosing the number of observations or people from a larger group (i.e., the TV population) to use in a HKTAM panel. The panel size should be large enough to give statistically valid results and accurate estimates of population parameters but small enough to be manageable and cost-effective. It is thus an art and a science.
Like other sample surveys, the panel size can be determined via mathematical formulas such as:

 
where n=sample size, t=the critical value from the t-distribution (e.g., t=1.96 for 95% confidence level), P=estimated TV ratings, M=acceptable sampling error.
Applying the above formula to HKTAM, with a sample of 2,700 individuals, the maximum sampling error for a rating point of 50% is ±1.9% at the 95% confidence level. For a rating point of, for example, 10%, the sampling error is ±1.1%, which means the true value can lie between 8.9% and 11.1%.
The formula shows that sample size n depends on the size of the acceptable sampling error M. As sample size increases, sampling error decreases, but the relationship between these two is not a simple linear relationship, i.e., doubling the sample size does not halve the sampling error. In other words, the reduction of sampling error brought about by a large increase in sample size (and thus cost) is not cost effective.
In addition, sample size determination also takes into account the individual behavioral differences of the target groups. If there are large differences in individual behavior, a larger sample size will generally be drawn; if the difference in individual behavior is small, a smaller sample size can be sufficient.
Another consideration for panel size is the level of investment available from the market to fund the service, and the time required for the set-up.
It is also important to mention that although panel size is a key consideration, there is no gain in building a big panel which is not representative of the viewing population. If the panel is biased, e.g., a particular demographic group is over sampled, the rating results will be distorted. Thus, CSM and Kantar have put extra efforts in ensuring the panel is representative to the TV population, starting from Establishment Survey, to panel selection, panel management, panel balance and quality control.
All in all, a large panel does not necessarily mean better and should not be an objective in itself. The objective has to be the accuracy, precision and stability of television ratings, which after all is the currency on which the industry relies.

Around 2,700 individuals aged 4 and above from 1,000 panel homes are being recruited to represent the TV population in Hong Kong. As discussed in Q.3 above, panel size determination depends on a number of factors, such as sampling method, acceptable sampling error, cost, and time.
Panel size of TV audience measurement services in different markets varies according to market situations and requirements. However, for the sake of comparison, we have listed the “household representation” of how many households a panel represents in some of the key markets , and how Hong Kong compares to these markets:

As shown in the above table, one panel household in HKTAM represents around 2,530 TV households in Hong Kong, whereas, in the UK, e.g., one panel home represents 4,200 households in the population. On the basis of this data alone, the representation of the HKTAM panel is higher than that in other markets listed above.
Also discussed in Q.3, representativeness of a sample depends not only on panel size, but also how the panel is selected. A large panel does not necessarily mean better if it is biased. The ultimate aim is to achieve accuracy, precision and stability of TV ratings, which after all is the currency on which the industry relies. To deliver that, every element of the measurement service has to work well and in practice that means properly investing in each. If they don’t, it’s actually possible for a larger panel to result in a smaller effective sample size.

No. The only eligibility criteria for joining the HKTAM panel is whether there is at least one working TV set in the household. Whether they watch TV and/or how much they watch TV are not a criterion for panel participation.

The PeopleMeter technology has been around for over 30 years and it remains the most commonly used technology to monitor TV viewing in markets with TV audience measurement. Kantar operates the most TV audience measurement services in over 60 markets worldwide, including Canada, the Netherlands, UK, etc., deploying over 70,000 PeopleMeters in more than 40,000 homes. CSM runs the world’s largest audience measurement panel in China, with around 56,000 panel homes. For other providers in markets like Australia, Singapore and the US, the PeopleMeter technology is also used.

The HKTAM Service, similar to most other audience measurement services worldwide, is governed by a joint industry committee called Founding Subscriber Committee (FSC). The FSC comprises representatives from the buy side of TV advertising - 10 media agencies from The Association of Accredited Advertising Agencies of Hong Kong (HK4As), and sell side of TV advertising – the three domestic licensed free TV broadcasters (HOY TV, TVB, ViuTV). This ensures a balance of interest and impartiality of the HKTAM Service.
CSM and Kantar are appointed as the official TV ratings provider of the current HKTAM Service, a joint decision made by the FSC through a tender process. Both CSM and Kantar are internationally recognized research companies on audience measurement and are independent from broadcasters. Both companies follow the international guidelines in setting up and implementing the ratings service. During the set-up stage and ongoing service, the FSC supervises and monitors the service quality to make sure it is up to international standards.
Furthermore, before the HKTAM Service was officially launched, an audit conducted by an independent auditor has been carried out to ensure the procedures and operating systems are properly executed as well as the confidentiality and privacy of the panel are kept. Only after CSM and Kantar have passed the independent audit did the FSC approve the launch of the HKTAM Service to the market. The Service is subject to additional audits if required.
Through all these measures, advertisers, media agencies, and media owners can have confidence in the integrity in the data as they use it for business decisions.

The Establishment Survey is the foundation of all TV audience measurement surveys, including HKTAM. It is a large-scale survey and serves 3 major purposes:
•    To provide universe estimates of the TV population in Hong Kong, and to determine the characteristics of the TV population that HKTAM will represent;
•    To determine the demographic and TV reception characteristics to which the HKTAM panel should conform;
•    To provide a pool of addresses from which qualified households, who have completed the Establishment Survey interview, will be randomly drawn and invited to become the HKTAM panel.
ES is conducted twice a year to monitor the dynamics of the competitive TV market.

It is NOT possible for people to volunteer themselves to participate in the Establishment Survey. Addresses for the Establishment Survey are drawn based on random sampling. This guarantees the randomness of the sample, which is vital in all research.

If you would like to make an appointment for the survey interview or have questions regarding the survey, please contact our Duty Manager at (852) 2111 8813 during office hours (10:00am-6:00pm, Monday-Friday, excluding public holidays).

As discussed in Q.1, at the core of any TV audience measurement is the TV panel. A panel is a representative sample of the population it intends to represent, such as by area, household size, gender, age, etc. Panels deliver a vital and trusted source of demographics and provide de-duplicated reach across platforms and devices, which are vital for advertisers to plan effective media campaigns.
TV audience measurement around the world often employs panel surveys, and the panel is relatively stable over time which allows continuous collection of viewing data from panelists 24/7/365, and over a long period of time. This will provide invaluable information on viewing trends, audience loyalty, engagement, and audience movement, etc.

At the core of any TV audience measurement is the TV panel. To give a reliable measurement service, a panel must be representative of the population from which it is selected. The achievement of a representative panel is first by employing a high-quality probability sample design and then by ensuring that the panel is, and remains, consistent with population values for a number of characteristics, known as panel controls. Panel controls are deployed to ensure that the panel remains representative of the Hong Kong TV population as the media environment changes (based on the latest results from Establishment Survey), and as homes leave to be replaced by newly recruited homes. In addition to that, the panelists are weighted at data processing stage to match the demographic characteristics of the TV population, to ensure full representation on those demographics.

The level of cooperation from panel members inevitably affects the quality of viewing data from an audience measurement panel. CSM and Kantar have put in place numerous rigorous and ongoing quality control procedures to ensure that the panel members follow the correct procedures.
Quality control procedures start with panel recruitment, where newly recruited households are trained on how to operate the equipment. Viewing data from these households are not included in the daily data release until all household members have completed the training.
On an ongoing basis, daily automatic quality checks are carried out on the collected data from panelists to detect any potential anomalies. If there is any doubt as to the integrity and quality of the data collected, the panel home will be contacted by CSM to verify whether the behavior is genuine (e.g., excessive viewing or nil viewing) and if re-education is required for the household in doubt. Follow-up action will be taken by CSM as appropriate.
In addition, CSM conducts coincidental surveys annually to assess the compliance performance of each individual panel member. The Founding Subscribers (FS) of HKTAM have set the accuracy rate at ±5% and CSM has met and exceeded this standard since the inception of the Service.

The panel home will remain part of the daily reporting panel, as this is a reflection of the market situation.

In general, panel members can stay on the panel for as long as they wish, provided that they continue to satisfy the demographic and quality control requirements. There is no maximum length of participation.

Panel members are not paid directly for joining the panel. Instead, CSM uses a ‘point’ system to incentivize households to join and remain on the panel. Points are awarded to panel households over the course of their membership. Panelists can use the points to redeem gift vouchers.

No. At the panel recruitment stage, each panel household is required to sign a consent letter which states that they cannot disclose or discuss their participation in the research, and should keep their participation in the panel confidential. If their identity is found to have been leaked, whether intentionally or unintentionally, they will be removed from the panel. This ensures that HKTAM remains as independent as possible and is not subject to the influence of other parties.

The privacy of HKTAM panel members is of paramount importance to CSM and Kantar, who have extensive experience in operating audience measurement services in different regions of the world.
Robust security measures are adopted to ensure that the identity of the panel households is kept strictly confidential. Within the data processing system, we have a number of security procedures in place to protect the identity of panel members. Only relevant staff can gain access to panel member information on a need-to-know basis. Access to the panel member database is controlled by the system administrator, who sets different data access rights depending on the role of the employee, and employees are required to change their login password regularly. In addition, the most sensitive personal data is encrypted for security reasons.
All data collected will be handled in accordance with the Personal Data (Privacy) Ordinance of Hong Kong. At result release, viewing data collected from panelists are aggregated and presented in groups (e.g., gender, age groups, etc.) to prevent information relating to individual persons or households from being revealed.
It is also important to mention that the FS (Founding Subscribers of HKTAM which comprises broadcasters and media agencies) does NOT have access to panel members’ personal information. This is to ensure the objectivity and independency of the HKTAM Service.

It is NOT possible for people to volunteer to join the HKTAM panel. All panelists are at the invitation of CSM, upon their completion of the Establishment Survey. Panel households are selected randomly, with controls in place to ensure that all sectors of the population are correctly represented. This ensures the randomness of the sample, which is vital in all research.

HKTAM covers viewing of TV channels available in Hong Kong, including free, pay/OTT, and satellite TV channels across all TV sets. The following types of viewing are covered by HKTAM:
•    Live viewing – Viewing of TV content during live broadcast.
•    As Live – Delayed viewing of TV content on the same day as the live broadcast.
•    7-day catch-up – Catch-up viewing on a date other than the same day as the live broadcast, covering up to 7 days.

Viewing activities on all TV sets in the panel households are recorded automatically in real time by Kantar’s PeopleMeter system that has been installed in the panel homes upon successful recruitment. A dedicated remote handset communicates with each PeopleMeter. Each household member (and their guests, if any) in the panel homes will declare their presence in a room, while a TV set is on, by pressing a dedicated button on the handset. The system records all such registrations made by individuals.
Throughout the day, the meter system collects all viewing information for the household and stores it in a memory. The meter in each panel household will automatically contact CSM’s processing center during a set time every night via the telecommunications network to transfer the collected viewing data in the PeopleMeter system for data processing.
At data processing stage, a complex audio-matching process (matching the audio captured from the panel homes to CSM’s reference sites of TV channels) is performed, to determine which TV channels were watched, and at what time.

No. HKTAM continues to monitor and report on the broadcast channels that are available in the Hong Kong market.

No, TV is watched by people of all ages. According to HKTAM data (Jan-May 2023), two-thirds of young people aged 4-34 watch TV on a weekly basis, averaging about 3 hours per week among viewers.

If the TV content from a broadcaster’s app on a smart TV or a connected device is streamed on the TV set, and that content can be matched back to the broadcast material stored in CSM’s audio reference sites within the last 7 days, those viewing will be reported in the channel’s viewing figures.

HKTAM can tell that the TV set is in use but at this stage, it does not report details on non-TV viewing activities.

HKTAM, like TV audience measurement services in most markets, is an in-home viewing service. However, if a person happens to visit a household that is on the HKTAM panel, his/her viewing will be recorded by providing his/her gender and age band via the PeopleMeter handset. This is called guest viewing which provides an estimate of viewing that takes place outside the viewers’ home, but in another residential home. Guest viewing is not included in the regular ratings data.

Since CSM and Kantar launched the current HKTAM Service on 1st January 2018, digital terrestrial TV (DTT) has been monitored and reported in HKTAM. With the Hong Kong Government’s decision to fully switch to DTT broadcasting on 1st December 2020, HKTAM immediately reflected the TV viewing behavior as Hong Kong entered the era of full DTT broadcasting.

The previous day’s TV viewing data is released to subscribers no later than 9:30am on the following business day.

HKTAM data is available by subscription. Interested parties please send an email to csmhk@csmhk.com.hk.