News & Views

With regret, Barb needs to announce a restatement of our published viewing figures for Her Majesty The Queen’s State Funeral on Monday September 19th 2022.

While our reporting of the overall audience across all channels was correct, the allocation of viewing to different channels contained some anomalies.

Barb’s ability to report which channel is being viewed relies, in part, on tracking the audio output of TV sets. Alongside this audio-matching technique, we use digital watermarks and service-information codes to provide certainty on which channels people are watching.

On September 19th, 52 channels broadcast the State Funeral, followed by the procession and Committal Service in Windsor. Never before have so many channels simultaneously broadcast the same programme with virtually no differentiation in audio for significant periods of time.

Having published viewing figures on September 20th, we identified some anomalies in how the audio-matching technique allocated viewing to different channels. These anomalies affected 2.4% of all the viewing that took place to coverage of The State Funeral, procession and Committal Service.

A fuller explanation of how this has happened is provided in the statement below from Kantar.

Amended data will be issued on Thursday October 6th.

We are sorry for the inconvenience this has caused.

———————————————

Statement from Kantar

The funeral of Queen Elizabeth II on Monday 19th September was both an extraordinary broadcasting event and an unprecedented challenge for audience measurement. The live broadcast comprised near-identical content on more than 50 channels. This content overlap, never before experienced in the modern broadcasting era, challenged the ‘audio matching’ element of our methodology. While the overall viewing figures calculated by Kantar and published by Barb were accurate, we recognise these measurement challenges resulted in some viewing misattribution between channels.

Kantar has conducted a detailed review of the audio matching data and will use the results of this analysis to adjust the channel-level attribution where relevant. We will reissue the data on Thursday 6th October.

The sophisticated TV audience measurement system Kantar has developed for Barb, together with the established rules of attribution, which we consistently applied, have three attribution steps. Kantar’s SNAP Watermark attribution, Sky SI/SK code attribution and audio matching attribution, which takes small sound samples from the content and matches it to a reference database of broadcast content. All broadcast content embedded with Kantar’s SNAP watermark has been attributed to the correct channel / channel-variant. All content broadcast on the Sky platform has been attributed to the correct channel / channel-variant using Sky SI/SK codes.

The third step, audio matching, which attributes the remaining viewing, has never before had to arbitrate across such an extraordinary volume of near-identical viewing statements, from so many different channels. It was in this step of the attribution hierarchy that the volume of datapoints analysed resulted in both a delay in publishing on Tuesday 20th September and some misattribution of viewing data. The review we have conducted will resolve this misattribution.

We are keenly aware of the responsibility we have in delivering accurate and timely data to the UK broadcasting community. We apologise to our partners for the inconvenience caused by the delay in publishing the daily viewing files for Monday 19th September, and the inaccuracies resulting from the audio matching challenges. We will use this experience to evaluate how our audio-matching technology can be further refined to ensure accurate attribution in the most challenging of broadcast environments.