Measurement across TV and tech platforms
And why this really matters to TV producers and the wider industry.
This post is about measurement, transparency, accountability and comparisons. It is another one of those posts which you might think a little irrelevant to you as a producer at the start, but I urge you to engage in this issue, as I think it is quite fundamental to the health of the content production and distribution industry - TV, streaming, direct-to-consumer, movie, creator. Frankly - the whole shebang.
I’ve covered the following:
YouTube sends a cease and desist letter to Barb following what they say is a breach of their agreement
How the story of YouTube having more reach than the BBC is probably based on an apples and oranges comparison
How all the various measurement companies have interesting data slices, but the lack of transparency and collaboration means we are missing an overall picture where like can be compared with like
A new report says that $65bn (or 8.5% of global digital ad spend) was lost to non-human activity.
In part this issue of measurement is most concerning to the wider advertising market, which for many in the TV production part of the world, have long been able to (largely) ignore. However, what happens in the ad market is going to have an even larger role to play in the success of your business.
This impact comes in two forms: firstly, the overall health of the content production industry depends on the corresponding health of the ad market. And secondly, for any creator or producer with direct-to-consumer or branded content operations, then what happens in the advertising marked directly impacts your business.
Why does all this matter? Well, as future of TV analyst Sofie Sue Rutgeerts summarised neatly on LinkedIn, the role of independent measurement is about good governance. As she wrote, there are three reasons it is vital:
Transparency - show up as you are if you want a share of revenues.
One common language - comparable figures are essential for trading and ad effectiveness - enabling optimisation, efficiency and limiting waste.
Market-wide adoption - no single actor should both set and mark the exam. If everyone measures, no one agrees.
Before I get to the main thrust of this week’s post, you might recall last week I did a video chat with Ricky Sutton all about Netflix’s 2025 earnings call. He has now published his version, and I’d really encourage you to read it - there is a whole wealth of insights into Netflix’s ad strategy which are worth your time:
YouTube and that cease and desist letter to Barb
So I was already planning on writing about the miasma we are swimming in when it comes to measurement, and then it was revealed that YouTube had sent a cease and desist letter to Barb and Kantar to get them to stop tracking individual channels which they said contravened the terms of their API service agreement (thanks to Jo Redfern):
FT: Google scuppers service comparing YouTube viewing with TV and streaming audiences
Google hands cease-and-desist letter over YouTube measurement
As a reminder, Barb tracks YouTube usage in various ways - they added them to the VOD tracking back in 2022, and last summer, they started a new initiative to track 200 individual self-selected YouTube channels (also worth noting YouTube opted out of being a Barb subscriber last summer).
This cease and desist letter is specifically about this new initiative to track these 200 YouTube channels. I wrote about this new initiative when the first set of results were published back in the summer, and for many, while it was welcome that Barb and Kantar was endeavouring to give more transparency around performance on the the platform, there were limitations in the data that everyone should be aware of (to avoid making a wrong conclusion):
By way of a quick summary, the limitations on this data were:
TV only, so no devices
A whole range of on platform content and behaviour was excluded e.g. Shorts, live streaming most notably (remember 75% of MrBeast’s content is Shorts, and live streaming is 30% of logged in views)
The tracked channels were self-selected by Barb and its partner organisation SeeViews, and they excluded all TV company channels (of which there are hundreds), but included some TV company owned and operated channels such as Bluey, Peppa Pig and Doctor Who
Not all content on a channel was included (but what was and wasn’t included wasn’t standard across all channels); in general, users watching old content wouldn’t necessarily have been counted.
It is this individual channel tracking that YouTube has objected to, where they said to The Media Leader:
YouTube has a long track record of providing access to third parties for research and reporting, and all third parties must respect the necessary Terms of Service and policies when using our API. While the vast majority of our partners, companies and creators adhere to these guidelines, we will take action when these terms are violated, as was the case here.
This cease and desist letter has caused many an eyebrow to be raised across the TV industry. For example, Lindsey Clay, CEO of Thinkbox:
It does seem odd that YouTube has spent so much effort trying to convince advertisers that it is TV and so gain the benefits of that reputation, but the moment there’s some TV-like scrutiny, it goes legal to avoid it. If they want to be treated like TV, they need to be transparent.
And here is Justin Lebbon and Ian Whittaker discussing this situation:
As Ian says, his natural response to this situation is why doesn’t YouTube want this information disclosed? He makes an analogy to when we go abroad we want to know what our money’s worth, so we have an exchange rate to do that calculation - here, we need something similar across the various content platforms. While Justin goes on to say while YouTube is happy sharing the big numbers of overall usage, they are less keen on granular breakdowns. He asks how much is toddler content, music videos, how much are 11 hour videos of fireplaces, niche creators, drama, or companion videos to keep dogs company at home alone?
I wrote about this conundrum a few weeks ago:
So this latest skirmish is against the backdrop of increasing industry frustrations about the relative lack of transparency when it comes to measurement - which is a core part of running a TV service - while simultaneously saying ‘hey, we’re TV’ as YouTube has been for some time now. For others, they think it is reasonable that YouTube enforces what they see as breaches to their service agreements.
As for why YouTube is unhappy about this specific initiative? Some have theorised it is because these Barb figures pulled back the curtain on how dominant channels for toddlers are (putting on 90 minutes of Peppa Pig on a loop for a 3-year-old might be great for the owners of Peppa Pig, but it might be less compelling to ad buyers that this equates to a TV-like comparison), or how small the reach is with particular audiences.
Others have speculated that individual creators might be concerned about how small their channel looked in the data (for example, much of the media reporting highlighted that the data showed MrBeast’s channel with fewer viewers than Quest's Celebrity Yorkshire Auction House. However the tracking excluded 75% of his traffic which is Shorts, a whole chunk of his archive, plus his four other channels).
Stepping beyond this specific issue with Barb; for others, the general lack of transparency is encouraging more questions to be asked about how accurate any of these numbers are in the first place.
As an example, this week, Entertainment Strategy Guy wrote the following (in relation to the relative poor performance of Dude Perfect’s movie performance at the box office):
YouTubers’ eye-popping stats often don’t translate off of YouTube. At some point, someone’s going to need to figure out this gigantic disconnect between social media stats and the inability to translate to any other medium. Something is off.
And it isn’t just YouTube. There are some creators on social complaining that since part of TikTok was separated and put in US ownership, their traffic has decreased. There are currently all sorts of platform service and stability issues, and also a whole bunch of users left the platform over the last few months of uncertainty. Which certainly may be part or the whole story, and once things normalise, it becomes clear this was a little temporary blip.
However, I’d like to take you back to February 2025 (again, in a post by Entertainment Strategy Guy), where he hypothesised that the reason Bytedance wasn’t easily giving up TikTok was because their user numbers were inflated. So now you can see TikTokkers openly wondering the same thing - were their numbers ever real in the first place?
Without independent verification, the risk is that distrust in the system proliferates.
As Ian and Justin note above, the position of the tech companies has been their data is neutral because it is scientifically based, all zeros and ones and therefore is unbiased. However, the counter argument is that this is a ‘just trust us’ situation, and involves these platforms marking their own homework.
And course, it is important to remember these platforms are answerable only to their shareholders, so aren’t in the business of releasing data that could negatively impact the company.
I’ll give you three further examples to ponder.
Barb’s SVOD data excludes a whole lot of broadcaster video and audio content, but includes all of YouTube, Twitch and TikTok
The YouTube is bigger than the BBC made for a lot of headlines and even more LinkedIn posts. Even Ted Sarandos mentioned it the Netflix 2025 earnings call. However, this conclusion isn’t the accurate picture - although I do appreciate this is probably a horse-has-bolted situation.
Barb tracks the major broadcasters linear TV and VOD services as well as streamers and YouTube on TVs, computers, tablets and mobiles. There is a quirk to be aware of, where some companies have not embedded the tracking tag into their platforms so out of home viewing via mobile connectivity or roaming wifi is excluded. So the BBC, ITV, C4, C5, Sky and UKTV have embedded the tag (and therefore any viewing away from the home wifi connection is counted), while it is understood that other companies like YouTube and Netflix have not done so for their own reasons - therefore the numbers reported exclude out of home viewing.
Even so, the YouTube tracking is all content and devices accessed via the home wifi connection. In contrast, the broadcaster tracking is only linear TV and VOD services, and not all the other video (or audio) content the broadcasters publish. And broadcasters publish a whole lot of content that bypasses their VOD systems.
Using the BBC as an example, the Barb report tracks linear TV and iPlayer views. However, it is likely to exclude the BBC News video views on the website and app, plus all the video views on bbc.co.uk (so any video content published outside of iPlayer). And if you wanted to be really punchy, well, YouTube includes all the podcast views on the platform, so why not incorporate all the traffic on BBC Sounds?
Here is a quick diagram of what I think is happening here (and BBC people, please correct me if wrong!):
Going further so as to really muddy the waters: YouTube carries radio stations too, so how about the BBC’s radio audience across the national and local stations? Never mind of course all the BBC’s video traffic on YouTube, plus the social video views off YouTube on other platforms. YouTube carries casual games which is likely to be included - but the BBC’s games output obviously hasn’t been counted.
How many video views or podcast listens are being missed? Well, I asked the BBC press office but they haven’t being able to help on that front. But to give even a vague sense:
BBC.co.uk’s total audience was 40.9m in November 2025 - but we don’t know much of that traffic was video, and how much (if any) is duplicated with iPlayer viewing
2.5bn plays of radio, music and podcasts January to November 2025 - however don’t know how much of that would count on BBC Sounds.
The key takeaway is that the YouTube platform is being monitored in entirety with no content excluded in that particular Barb report, while the BBC’s platform is only partially being monitored, as output is excluded across News, BBC.co.uk and Sounds.
The same will apply to all the other broadcasters, the majority (if not all) will have video content published via their VOD systems, and video content published independently. So for example, the Barb data will be tracking Channel 4’s All4 VOD content, but probably isn’t tracking the non-VOD views such as the content on Channel 4 News’ website.
You can see how this ends up being visualised in the Ofcom Media nations reports (for example, 2024) is below - it says total video viewing, but it isn’t - it is a subset for the broadcasters, and complete for the tech platforms.
It is important to say here that the audience for Barb’s report are broadcasters, streamers, commissioners, schedulers, planners, marketers and ad sales. The way this tracker functions is essential for these people to do their jobs, and they all have the full knowledge of what it means and what it doesn’t.
However, it has now been used to jump to conclusions about total audience reach, without any of these caveats in place, and paint a picture that is more favourable to the tech platforms and arguably unfair to the broadcasters.
So is this a classic apples and oranges tale, and you can see how a particular data set for a specific B2B audience became a great headline and as a result, has now been routinely repeated as fact.
Different data sets and slices paint different pictures
Following on from the above point. We are in a phase where all sorts of analytics and data insights organisations are tracking all sorts of different dimensions, which are interesting and useful to their clients and stakeholders but needs a whole host of digging before realising what has been excluded.
This is why time and again, we see a story running around the block before there is a chance for someone who has looked under the hood to add a ‘well, actually...’ caveat to what this data tells us, and what it doesn’t.
Here are just a couple of examples of attempts by organisations to get under the hood of what is happening on YouTube and TV audience behaviours.
This week, eMarketer published this by analyst Bill Fisher, showing that in the UK, while YouTube reaches a similar number of viewers to TV, it trails significantly in terms of time spent.
In a way, this isn’t surprisingly as YouTube is many things to people - so a quick visit to watch an explainer video on how to fix your washing machine versus a long form entertainment destination, and a huge amount in between.
This is a great example where granular transparency from the platform would be so helpful - how many of these viewers are watching ‘how to’ videos? How many are watching 2+ hour live stream gaming walk through or listening to music during the day time? How many are watching TV-like content? How much is Shorts? How much is podcasting? How much live radio? Just how popular are 10 hour fireplace videos? And so on.
Another example. UKOM (The UK’s Online Measurement Body) endorsed the following data from Ipsos Iris, which last November showed that on devices other than TV (so desktop, tablet and mobiles), at home and out of home, the UK’s search for 15+ includes music labels and aggregators as well as all the main media companies of the BBC, ITV, Sky, Channel 4 and News UK. (Although I’m not clear which content owners have been included and which haven’t).
As Ian Dowds, CEO of UKOM commented:
These organisations [the media companies'] don’t seem to be losing. They’re all fighting hard on multiple fronts…and seem to me to be making a good fist of it.
Meanwhile, ViewersLogic shared this table, where they split UK YouTube channel traffic by number of views and number of viewers, aggregating the media companies channels together (where they have the channel brand in the name - so Channel 4 Entertainment - but not including specific shows, such as Bluey or Doctor Who):
What does this show? That channels with a high volume of Shorts, or highly repetitious content (so music videos, and videos for toddlers) rack up views but probably with a smaller audience reach. While reaching lots of viewers is where traditional media’s brand and reputation comes to the fore. Although Yee Yee Life at number 2 was a surprise:
As interesting as all of these data points are, they are even further illustration of the frustrating lack of clarity coming from the platforms themselves, so all of us are spending a lot of time and effort trying to piece together bits of data to try to come up with a clearer picture.
The level of ad revenue lost to fraud
I’ve written before on the issue of real versus fake traffic online, and yet sometimes we can forget that this is live and ongoing issue that has real potential to blow up over the coming years.
Last week, there was a report by Lunio that said $63bn of ad spend was lost to ‘invalid ad traffic’ in 2025 - meaning any click, conversion or web activity that is non-human. That is out of a total ad spend budget of $740bn.
This isn’t a single instance, rather instead there is growing unease coming from the ad industry, for example:
Omar Oakes post ‘Stories That Matter: Are We the Bad Guys?’ is worth reading on the above industry concerns.
There was also that story from last year that reported 10% of Meta’s earnings are from fraudulent ads:
What does ‘invalid ad traffic’ mean? It can cover a whole host of sins, but by way of a generalised shortcut, I’d suggest picturing a bot phone farm it is mentioned. Indeed, I’ve viewed so many on TikTok that my feed is now full of bot farm operators offering me their services.
What it means in practice is companies - via their marketers or ad agencies - are paying for advertising that is of no value as it doesn’t get presented to an actual real life user.
Last year, Polygraph (a software company to help marketers spot when their budgets are being siphoned off by click fraud) shared data that there 3 out of 4 interactions with adverts on TikTok is by a bot not a real human.
There are those across the ad market sounding the alarm about ad fraud and lost revenues - with a further suggestion it is going to get worse once AI is layered in on top.
Dr. Augustine Fou has been tracking this issue for years, and has also recently made the point that while the $63bn number mentioned above is ‘directionally correct’, it isn’t the complete picture due - again - to that lack of transparency in the ecosystem.
Lunio’s country estimates are below, and Augustine Fou’s point is that simply multiplying the click fraud percentage by the ad spend by country isn’t the right way to get to an estimate.
We are in this catch 22 where we have more data that we’ve ever had before; the internet creating whole new routes to customers all over the world for businesses; indeed businesses of all shapes and sizes are emerging about to use the myriad of platforms to reach potential consumers in every which way.
And yet it is increasingly seemingly like a tangle, especially when you layer in the lack of independent universal measurement. To give a flavour of how complicated this world is, here is Luma Partners’ map of the video landscape in 2025 (this isn’t the display ad market - that has a whole other diagram).
So as this world is getting increasingly complex, there are some saying that even more problems are entering the system and that signals are being wrongly interpreted.
There was a great session at VOD Professional’s OTT conference this week where ITV’s director of advanced advertising Rhys McLachlan talked about how sanitised language such as ‘invalid ad traffic’ is used to describe fraudulent behaviour, and how collective blind eyes are being turned to behaviours that have real consequences for well established businesses like ITV. His overall point was that an easy ride is being given to the platforms in comparison to the much harder position our public service broadcasters are dealt.
Although the ad industry is far larger, these noises do remind me of the premium rate telephone scandal back in 2007. This was where TV broadcasters made lots of money from phone competitions where each called was charged to the user, but in reality they had no way of winning as it was prerecorded, or they kept taking calls even after a winner was selected. The similarity is when something considered the normal way of doing things suddenly stops being acceptable; where tolerated levels of behaviour become beyond the pale, and everything is upended.
Stepping back more widely to the broader themes of measurement, transparency, accuracy; the goal of being able to make accurate comparisons between media companies, platforms and content.
Are the issues I’ve explored here just tricky teething problems that will be teased out over time, and are just a reflection of where we are at in this process of media convergence? Will we be sitting here in 10 years time, looking back to see how patchy our measurement culture was before clearer standards emerged, or how prevalent ad fraud was before it got stamped out?
This question loops back to advertisers really, which are businesses with balance sheets and bottom lines. Much of business is about confidence. And while there has long been pockets of skepticism around particular areas of measurement or metrics, in general, there was confidence overall that things were as they seem.
There are flags going up all over the place with concerns about the veracity of measurement across our digital ecosystem, the comparisons being made, the uneven playing field, the lack of transparency, and crucially, the issue of ad fraud; do these flags constitute a wobble in confidence in the ecosystem? Or are they just the creaking and groaning of an industry going through a profound period of change?
If they do constitute a genuine wobble in confidence in the ecosystem, will it be the brands who start to pull back the curtain on their digital advertising strategy, and wonder if the results they see in their balance sheets tally up with what the ad metrics are saying?
That $65bn fraud stat you flagged Jen really drives home the measurement chaos. Your YouTube-Barb breakdown shows why TV ad trust is crumbling fast.
Quick fraud deep dive. That invalid traffic isn't random bots anymore. Coordinated farms now loop CTV streams and YouTube fireplaces for fake hours watched. Toss in AI scripting human-like sessions? Tests show 30-40% non-human inflation in TV/digital hybrids. Skewed reach kills Barb-style comps and drains real ad budgets.
Game changer is hybrid signal scrubbing. Real-time device graphing plus behavioral baselines. Catches scripted fakes before they hit metrics. Hands advertisers' verifiable clean views across platforms.
wow very valuable. am curious. if u had to select services that measure streaming performance as a complement to the territories nielsen covers, which would u pick