One response to Channel 4's Mirror on the Industry report finding that black people are represented in 51% of commercials on UK TV is to say that there're multiple people in each advertisement, so this doesn't show a problem (I can't find the original tweet about this).
There're many things problematic about the Mirror on the Industry report. For example, it track the representation of LGBTQIA+ and Disabled People. Yet, what does a LGBTQIA+ or Disabled Person look like? Katy Montgomerie, a LGBTQ activist, says that you can't always tell if someone is trans (and gives examples of trans people who were outed - i.e. people couldn't tell before they were outed). And we have it on good authority from the UK Government that non-visible disabilities are a thing, and that we cannot ignore people with them. Indeed, Usdaw, "one of Britain's largest trade unions with over 360,000 members nationwide", informs us that 80% of disabilities in the UK cannot be seen. And representing every single "minority" in the tone and detail that the report demands is a huge headache (especially given that the report critiques montage-style ads - when montage-style ads are the only way to represent more groups). This is enough to put almost anyone off the task of pushing DEI in ads (which the report trumpets the importance of).
But let us go back to the seeming high representation of black people in UK ads.
There is a point in this - if, for the sake of argument, every ad had 100 people in it, and in half of adverts, 1 person was black, we could report that 50% of ads featured black people, but black representation would still be pretty poor (0.5%).
This suggests that to game this diversity metric, the solution is to have ads featuring as many people as possible, with each "minority" ticking off multiple boxes: for example, if you have an old, fat, black pregnant working-class transman in a wheelchair, that would tick seven boxes at once (getting it to, as the report demands, have a speaking role, be a lead character and not be in a 'stereotypical' job, whatever that might be, would be more complicated).
There is also a problem in the other direction - if one ad featured 100 black people in it, but 99 other ads had no one in them, we would erroneously report that only 1% of ads featured black people, when black people really made up 100% of people in ads.
A more useful metric, then, than the percentage of advertisements with black people (or whatever "minority" you want to measure) is the percentage of people in ads who are black.
Unfortunately, the report does not disclose this, but we do get a proxy answer on Page 34, which reports that 19% of characters portrayed as part of a family in 2024 were black. Finding primary sources for the black percentage of the UK is difficult - 4.0% of England and Wales was black in the 2021 Census, but I am unable to find primary sources reporting combined numbers for the whole of the United Kingdom. Wikipedia reports it as 3.7%, which is not unexpected given the populations of Scotland and Northern Ireland, so we can use that number. So we can see that, at least as part of a family, black people are significantly over-represented.
To confirm the results, let us try to look at the broader universe of ads, and do some calculations about the probability that pure chance could've resulted in the representation we see.
Channel 4's Mirror on the Industry looks at 1,000 UK TV ads from the previous year with highest adult impact. Unfortunately I do not have the manpower to look at 1,000 ads (if you want to fund me, we can discuss this separately). Nor do they provide details on how they select these, much less a list of the 1,000, though the research partner Tapestry does give a few details on how they code the ads.
I was also unable to find a list of the most viewed UK ads, only the "best" ones or "top" ones in curated lists, or the most viewed Christmas ads, or people's "favourite" ones.
The best source I could find was 2024’s 10 most-viewed ads on The Drum.
Here're the results of my count for each ad:
1) Heinz ‘Ketchup Insurance’ by FP7 McCann - 0 people (it's a collection of social media posts, a mock insurance policy and some graphics)
2) Cadbury ‘Yours for 200 Years’ by VCCP - 46 people (more than half of this is set in the past, but we know that historical anachronism is not a problem for diversity)
3) Apple ‘Custom Memory Movies’ by TBWA\Media Arts Lab, Los Angeles and Sydney (the embed wasn't working and TikTok didn't let me play the video even after I signed up, but I found another version) - 4 people
4) KFC ‘Tower Burger’ by Mother - 0 people (we just saw the chicken sandwich)
5) Tesco ‘Icons’ by BBH - 0 people (this was a static billboard and not a video ad)
6) Target ‘Born to be Kris’ by Target Creative - 1 person
7) Haribo ‘Kids’ Voices’ by Quiet Storm - 13 people
8) Beats by Dre ‘Beats’ by Mirimar - 15 people
9) Volvo ‘Meet the New Volvo’ by Hoyte Van Hoytema - 26+ (a lot of characters were blurry and/or the shots lasted a very short time)
10) Decathlon ‘Outage’ by Rethink - 0 people (this was a static billboard and not a video ad)
So in 8 ads, we had 101 people, for an average of 13 people an ad. Note that 2 of the ads featured 0 people, so they could not have featured a "minority". This doesn't stop these ads being chucked in the "no representation" box, since the assumption is that if an ad doesn't feature a "minority", it must feature straight white cismen.
Another huge problem with the counting was that it was often unclear how many distinct individuals were in each ad, as characters often were not identifiable. This ties into a bigger problem with this diversity audit: I couldn't even tell the race of many of the characters in the ads (gender was usually identifiable, but not always). So besides non-visible minority status; I have mentioned LGBTQ+ and disabled people earlier, but there're also lots of problems identifying ad characters who are 8 weeks pregnant, high functioning autistic people, working class etc - presumably the solution is to be even more heavy-handed and announce all the characters' attributes because Representation is so important.
Be that as it may, with these baseline numbers, we can now calculate what the probability of a randomly sampled ad (with randomly sampled casting) having a black person would be:
(1 - 0.963^13) = 38.7%
To unpack the maths, this calculates the probability that all 13 people in a UK ad would be non-black, assuming that the casting sampling was random from the general population. Subtracting this from 100% gets you the probability that at least 1 of the 13 would be black.
So we can indeed say that black people are over-represented in UK ads. In reality, given that
Also, given the issues with coding I have previously mentioned, the true over-sampling is likely more (unless we assume that all characters whose race cannot be clearly identified are white or another non-black minority which, while plausible, likely does not apply 100% of the time since the ad coders may not be that perceptive).
