using website header

Connected: Display Connected: Media Landscape Connected: Regional Connected: AV Consumer Surveys Connected: Direct LinkedIn LinkedIn logo icon Twitter Twitter logo icon Youtube Youtube logo icon Flickr Flickr logo icon Instagram Instagram logo icon Mail Mail icon Down arrow
Dominic Mills 

Number wanging the Facebook way // Grade A toadying at Ad Week

Number wanging the Facebook way // Grade A toadying at Ad Week

Facebook is no stranger to fudging numbers, and Dominic Mills has caught the social media giant at it again. Plus: Matt Scheckner's undying love for his biggest funder.

Numberwang was a Mitchell and Webb game-show parody which involved the pointless calling out of random numbers.

Facebook has lots of previous experience in this area, and somewhat laughably claims that its focus on transparency means advertisers have nothing to worry about. The phrase ‘lies, damned lies and statistics’ doesn’t really do this justice.

This week’s column has nothing to do with vastly inflated video measures, but instead involves some classic number wanging by Facebook in the area of reach.

My attention was drawn when the issue was raised a few weeks ago at Mediatel’s Future of Media Research conference.

A little simple digging reveals the scale of the problem. Facebook looks to be over-claiming reach in a few key basic demographics by 10-15%.

Take the 15-24 age group. Using Facebook’s self-serve advertising tool, it claims a potential reach of nine million people in the UK (comprising England, Scotland and Wales).

Here’s the page below:

Putting aside the convenient round number, it all looks so straightforward, doesn’t it? Except that there aren’t 9m 15-24 year-olds in Great Britain (i.e. including Northern Ireland), let alone in the UK.

At least not according to the ONS, whose mid-2015 data gives a figure of 8.118m, including Northern Ireland. Or those figures from the wonderfully named JICPOPS - who even knew such a body existed? - which gives us 8.24m for the four home nations.

Given Facebook’s previous misdemeanours with numbers, I know whose estimates I prefer.

I make that an over-claim of around 10%, a bit more if you factor in Northern Ireland. Similar, er, alternative facts, can be found in other broad demographics: for 25-34s, Facebook claims a 10m reach UK versus an ONS GB estimate of 8.822m and JICPOPS 8.83m; and for 35-44 year-olds, Facebook claims 8.6m UK reach versus and ONS GB figure of 8.4m and JICPOPS’ 8.37m.

How might this happen? It’s not difficult to work it out: some people will have multiple Facebook accounts; others will have lied (up or down) about their age.

And, in the grand scheme of things, this kind of statistical sophistry is small beer - certainly compared to ad fraud, brand safety and so on.

Indeed, for performance advertisers it is a minor irritation. But not for those in display or focusing on reach - as P&G does with Facebook - it is a more serious error.

And for other media owners who compete with Facebook in display, and who are held to higher standards of account, it must be deeply frustrating. Even if they thought they could get away with it, what would Group M say if they over-claimed reach by 10-15%?

But what really bothers me is it’s such a stupid mistake to make - especially as it’s so easily challenged - and one that I think gets to the heart of Facebook’s attitude.

Curiously, I am not the first to raise the issue. This blog by James Cridland goes back five years when the over-claim was even bigger.

Some obvious questions raise themselves:

- Who’s responsible for publishing and maintaining Facebook’s reach numbers?
- How senior are they?
- Who checks their work?
- How high a priority do reach stats have among Facebook’s ad teams?
- Why don’t they check their figures against official estimates?
- Is it hubris - they’re Facebook numbers, so they can’t be wrong - ?
- And, more worrying, why do advertisers and agencies believe them?

I raised the claimed reach numbers and, just before deadline, here is Facebook’s response:

"Facebook’s age data is from self-reported data. At times, the age ranges may not align perfectly with census or partner estimates given the difference between census response rates (modelled data) versus self-reported.

"We have a range of independent third parties who regular [sic] measure the accuracy of our targeting and it is found to be highly accurate. We support age targeting/minimum age limiting based on the information provided.

"Here are a couple of useful link[s] to look at how we are partnering with third parties for audience verification of our targeting data:


"Facebook’s targeting currently includes 8.3m 16-24s within its estimates."

Make of that what you will.

Grade A toadying at Ad Week Europe

What’s not to like about Ad Week, especially when, with due serendipity, it coincides with the Google/YouTube firestorm?

Google, which sponsors the arse off Ad Week, therefore found itself front and centre stage in every conceivable way. But what does an event organiser do when it finds its main sugar daddy suddenly friendless and under fire?

Thus, in an example of heroic, Grade A toadying, up steps Ad Week head honcho Matt Scheckner (pictured). Addressing the audience ahead of a set-piece between Unilever CMO Keith Weed and Google’s Matt Brittin, he declared his undying love and respect for his biggest funder.

Sounding not unlike Trump bromancing Vladimir Putin, he said: “They’re [Google] vital players in the broader industry eco-system to their tremendous, tremendous credit. They tackle everything head on with such vigour, such commitment, and I know that the commitment to doing the right thing is genuine.” To borrow from Trump, Google - in Scheckner’s view - are ‘good hombres’.

And what form does this ‘commitment’ take? Well, it doesn’t involve a pro-active approach to taking down hate, racist, or extremist videos, despite the best efforts during the session by the British press (questions from The Times, the Mail, ITV and the FT) to commit Brittin to such effort.

Instead, ad nauseam, he repeated the Google mantra: a three-pronged approach, involving the Google community, technology and the Google team itself, the latter acting, as far as I can tell, in a purely reactive way. Not a whole-hearted commitment, then.

You can watch the video of the session here. Scheckner is at the beginning, the press hounds from 20 minutes on.

Entirely separately, it seems to me that Scheckner’s threat on the opening day to move AdWeek from London to Amsterdam - using Brexit as a fig leaf - was ill-judged.

Telling delegates and sponsors who have just forked out significant sums of cash that you might take your ball away is, at the very least, a slap in the face.

Scheckner deserves huge credit for his vision and drive in setting up Ad Week (see, I can toady too, but only at B level), but he would have been better advised to stay schtum.

I wonder how much the threat of quitting London has to do with the fact that some previous Ad Week sponsors - ITV and Sky, for example - have kept their hands in their pockets this time round.

If the big UK media owner sponsors drop out, then Ad Week becomes even more reliant on the likes of Google. Given the current climate, they’d probably be very happy to get out of London.

Leave a comment

Thank you for your comment - a copy has now been sent to the Mediatel Newsline team who will review it shortly. Please note that the editor may edit your comment before publication.

BenThomas, Planner, Earnest on 15 Jun 2017
“If someone gave me the tools I could knock a building down... give me all the tools in the world and I wouldn't be able to build it again.

It's much easier to pick things apart than it is to find a solution. Everyone knows data isn't perfect. It never is, it's never going to be. You can take steps to clean it and take it with a pinch of salt with the correct measurement plans.

What do you believe to be the solution - what metrics should we use?”
NigelJacklin, MD, Think Media Consultancy on 30 Mar 2017
“The facebook claim (potential reach of 9 million people in the UK) is careless. As you state, people with more accounts could be counted twice (but they are not two people) and if, when you were 11 you pretended you were 13 (to legally sign up to Facebook), given the age of Facebook, this would also lead to overstimates. The other careless error is calling England, Scotland and Wales the United's GB not UK. Having spoken to a few people under 24 we have identified several reasons for having two accounts or dual identities and so assume this is commonplace. On the other hand there could be a million people living her illegally...but I suspect they tend to be a bit older.”

Media Jobs