News Analysis: Facebook v Census - out for the count?

Charlie McKelvey, contributing editor, DataIQ

Proof of digital marketing's world domination is never hard to find - hardly a day goes by without a new report of its ascendancy. Proof of its accountability is slightly more difficult to come by, however, and despite the best efforts of Procter & Gamble chief brand officer Marc Pritchard to push for accountability, there are still many issues to tackle before the industry can shake off its "smoke and mirrors" image.

Not that Pritchard is alone. According to a recent study by the World Federation of Advertisers, concerns over the online ad industry’s overall level of measurement are growing, with the majority of the world’s largest advertisers - with a combined annual adspend of more than $80 billion (£73.3 billion) - “dissatisfied” with current standards.

Enter Facebook, second only to Google in the amount of advertising spend it attracts, and recent reports that it has been exaggerating its audience reach across the world. In the UK alone, Facebook claimed to reach 12.2 million adults aged between 20 and 29, even though, according to the Census, there are just 8.76 million people in this demographic.

In its defence, Facebook said the tool is an estimator for campaign planning, rather than something that quantifies officially how many people a business can reach in a given campaign, and is based on a number of factors, including user behaviours, user demographics, location data from devices, and other factors.

In a statement, the company said: “They are designed to estimate how many people in a given area are eligible to see an ad a business might run. They are not designed to match population or census estimates.”

But the issue has raised questions about whether the UK Census, launched in 1801 under the Conservative government of William Pitt the Younger and carried out by the Office of National Statistics (ONS) once a decade, is still fit for purpose.

Although marketers and other organisations have to wait at least two years until they get their hands on the Census data, it not only provides crucial information on consumer trends and demographics, as well as social data, it forms the basis of many new products. Census statistics are also used to allocate public money for local services, including education, housing, transport and health. Over £100 billion per year of public money is distributed nationally using population estimates.

Dr Tim Drye, DUGFor Dr Tim Drye, director at Data Analysts User Group (DUG) and managing director of DataTalk, it is symptomatic of the current stampede towards digital that Facebook thinks it can ride roughshod over official Census data. He said: "It's a classic example of where Facebook has been over-claiming and goes to the heart of the issues over digital marketing measurement. All the time, Facebook is attracting millions of pounds of advertising spend, it seems to do what it likes.” 

“Companies look at the wrong analysis to carry out their targeting - they look at what they want to see.”

“There is a herd mentality among marketers that they have to follow everyone else. All too often, companies look at the wrong analysis to carry out their targeting - they look at what they want to see, but they should be looking outside that group,” he added.

And Blueberry Wave chief operating officer, Steve Mattey, reckons the real reason behind the  furore is that Facebook has become a very powerful and potent marketing tool where vast sums are being spent by advertisers. “No-one wants to look like they've been a fool,” he said.

"It's all too easy to get sucked along with what seems a great idea. Sometimes, it needs someone to just say, 'does this make sense?’ The rapid fag packet calculation on whether a plan stacks up or not sorts the wheat from the chaff. Perhaps the learning here is that a little more due diligence over precisely what you are buying wouldn't go amiss, and robust measurement and evaluation would certainly highlight such large discrepancies,” said Mattey.

Mattey reckons that Facebook and similar channels are still unproven in delivering return on investment over a long period to brands. The diligence and learnings that have been the bedrock of direct marketing for decades have yet to be applied properly in these channels. So, perhaps, this is all for the good in the end, he adds. "At the very least, the first question I would ask Facebook before parting with cash would be, 'so, how, exactly, do you arrive at this audience number?’"

Not that everyone is in agreement, however, as Zhewei Zhang, assistant professor at Warwick Business School, explained: "I don't think advertisers should worry too much about this difference. It is quite understandable that there is a such gap between Census data and Facebook's estimate, which is basically the gap between the number of residents and the number of potential ad-viewers."

Zhang points out that the Government has a stricter definition on who "lives in" the country, compared with Facebook's definition on who is "in" the country. Facebook's estimate cares more about who can view its advertisements.

"The estimated data from Facebook is more of a PR strategy.”

He believes the difference can come from factors such as immigration status, whether the viewer is a resident or visitor, legal or illegal immigrant, and also age, whether it is from official birth records or self-reported, especially as it is not uncommon for juveniles to raise their true age in order to access content with an age restriction.

Zang added: "The estimated data from Facebook is more of a PR strategy to show its potential total number of reach. For any advertiser, the number of effective viewers is much more important and Facebook claims it can provide more accurate targeting and better results than traditional sells."

And Mattey believes that there is an element of "people in glass houses" about the claims, noting that, "the loudest wailing and gnashing of teeth came from the press and some sections of the marketing industry. I think first those that are complaining most loudly need to have a hard think about what they offer. Newspapers publishing a story that hasn't been fully checked out? Heaven forfend. Marketing which makes a bold claim using facts which best suit their argument? Knock me down with a feather."

His view strikes a chord with Pete Gatenby, client services director at mobile marketing consultancy B60, who said: "I don't think Facebook advertisers should be any more worried than they should be about other marketing medium claims. If anything, I would be less worried. Facebook provides hugely impressive analytics in relation to ad performance that allow marketers to test, measure and improve, something that more traditional media certainly don’t."

“It suits Facebook to have as many accounts as possible to maximise reach (and revenue).”

Caroline Kimber, Stack AgencyFor Caroline Kimber, data strategy director at Stack Agency, the issue highlights clearly one of the fundamental challenges with digital data - how to get a single customer view. She points out that, as Facebook measures accounts rather than individuals, it is not atypical to see an individual with a regular account, an old account that is not used but has never been shut down, and  further accounts for, say, a small business that they run or a fansite they have set up for a brand or celebrity that they admire.

Kimber adds: "Of course, it suits Facebook to have as many accounts as possible to maximise reach (and revenue) - there's no benefit to them in linking accounts back to one individual to get a single customer view."

So, does the Census fair any better? The ONS has certainly recognised that its paper-based survey is far from perfect and has been investigating other methods of data capture. Last year, the organisation commissioned four pilot schemes due to report back in 2023 to make the Census more relevant to modern life.

It is hoped that finding new sources of data, including information held by Government departments such as the Department of Work and Pensions and HMRC, as well as from mobile phone users and electricity smart meters, will improve the information provided. At the time, Jane Naylor, head of the ONS big data project, insisted the plans were not just about cutting back on the huge cost of running the survey, despite the last Census in 2011 costing £625 million and being the most expensive project the ONS has undertaken. She said: "It’s about updating and modernising the output that we can produce, keeping relevant and meeting our user needs.”

Kimber agrees that the current format needs to evolve into something more agile that can keep pace with the rapid changes in society, but added: "I think that the Census continues to have an important role, it's a valuable baseline."

And it is the depth of detail, as well as the accuracy of the data at the time it is gathered, that makes it a powerful tool for analysts, according to Drye. "No-one can analyse just from their own internal data, you need something to benchmark it against and that is where the Census data, and other robust data-sets, come into their own," he explained. "Online surveys only paint half the picture. You can't under-estimate the power of interviews with 'real people'."

Ultimately, then, when it comes to building a clear picture of UK consumers, it seems that the humble pen and paper - and a 216-year-old survey - still has the edge over anything that the digital age can provide.