Faker Data
02 November 2026
Comparisons of different video channels are commonplace on social media. Whether based on ‘Reach’, ‘Attention’ or ‘Cost-per-Thousand’, they aim to satisfy a need for a ranking. They’re the listicles of the media world, and about as informative. As one commenter had it: ’Why are we comparing shoelaces to trees?’
We have become so obsessed with numbers that we’ve forgotten how to ask questions. We’ve misplaced our critical judgement.
Imagine yourself as a CMO. In amongst a thousand other things to do that morning you need to make a decision on whether to greenlight an ad. Will it work? Do I like it enough to defend it?
Some are quite comfortable making a call based on informed judgement.
Most are not.
Imagine being the Sony CMO explaining this Bravia ad of 2005 to his colleagues. ‘We’re going to take 250,000 coloured balls to the top of a hill in San Francisco and drop them. We’ll film what happens in one take. It will make us famous.’
I don’t know how much data the agency, Fallon, had to justify their idea, but I would be prepared to bet that this was a judgement call. A brave and ultimately successful one.
Judgement plus data. The data says … the data shows … surveys indicate. A balance between indicative facts and judgement – ranging from 100% one way to 100% the other.
Data is safe; data is inanimate, unquestionable, unemotional, secure. Data can be a killer of bold ideas.
There is an irony here. We think of data as a rock; solid, reliable. In fact, despite appearances, the rock is often made of polystyrene. It would be foolish to rely on it without asking the simple questions, like: ‘Is this actually a rock?’
To over-simplify, we make the mistake of assuming the same word used by different people in the same context means the same thing.
Take the word ‘see’. How many see at least one ad within a campaign. Usually expressed as ‘reach’.
What does ‘see’ mean?
Facebook, TikTok, YouTube, BARB all define ‘see’ differently. In some, the seeing starts from the moment the user clicks ‘play’ regardless of how much is ‘played’. So, in theory you can be said to have seen something you haven’t in any sense ‘seen’.
Others apply a minimum of 3 minutes to the service within which the ad appears.
These are different things. What you thought of as a solid, definable verb has become a debate.
What about the absolute number said to have an ‘opportunity to see’ something via a channel’s reach?
In 2017 FB claimed the channel’s ad tool could reach 41 million Americans aged 18-24. At the time, the US Census had the population of 18-24s at 31 million. Similar discrepancies occurred in other markets too.
Facebook were claiming to reach more people than existed.
So what? ‘seeing’ an ad in print or on OOH was always defined differently to ‘seeing’ an ad on TV, but the examples above are all for an ad playing out on a screen. Why are they different?
Definitions do matter – and not just to us geeks. We should be confident in the comparisons we make.
Maybe the best audience measurement experts meeting this week at the asi event in Venice could advise? Isn’t this rather fundamental? I mean, if we can’t even agree what ‘see’ means…
Last month Pesach Lattin’s newsletter ADOTAT reported that across DSPs the percentage of ad traffic that is identified as fraudulent runs at around 1% to 3%. And yet Anura, a business that detects and reports on ad fraud sets the level at around 50%.
As ADOTAT correctly points out, these are measures of different things. Even if they’re both called ‘fraud’.
The 1% – 3% relates to the fraud picked up by the DSPs as ‘general invalid traffic’; it’s not subject to independent verification. Further it excludes the sophisticated fraudulent stuff. The 50% on the other hand includes everything fraudulent.
I can testify to the differences Pesach identifies from a real-life, client assignment – as can Augustine Fou who has been at this fraud detection game longer than anyone.
As a marketer doing his or her best to understand and quantify the true audience to a campaign who are you to believe?
Your media agency? Do they accept adding apples to oranges? Do they act on the low fraud numbers produced by a supplier with whom the agency may enjoy a relationship, and whose data suggests the agency is on top of the issue?
Or do they favour an independent specialist whose conclusions suggest that maybe the agency isn’t doing quite the job you might expect?
Choices, choices.
Skilful marketers will continue to back their instincts, amongst which is an ability to spot and use real rocks.
Others, increasingly the majority of advertisers, will rely not on human skills but on the stability of something sold them by polystyrene rock salesmen.
Shoelaces and trees, people. Shoelaces and trees.