When the victims of famed imposter Christian Gerhartsreiter were asked how they could fall for the man’s claim that he was a member of the illustrious Rockefeller family, the response was generally some variation of ‘because that’s what he told me’. The contact lenses given to us by hindsight make this rationale seem ridiculous now, but it’s a reminder that the default setting for humans is one of trust.
Something very similar has happened in digital advertising. Some years back, search and display introduced themselves as a pair of sophisticated gentlemen, who had arrived to change things around town. They came bearing briefcases filled with the finest assortment of tools, designed to cut away the fluff and give marketers greater efficiency.
Buoyed by well-crafted press releases, news of their arrival travelled quickly and before long every advertising journalist was writing about the benefits of targeting customers through digital.
At first, few mistrusted the claims that were being made. And why would they? Digital was, after all, presented as the first truly measurable media channel. Every action could, in theory at least, be traced as online users wandered from one page to the next, providing valuable data perfectly suited for application to online marketing campaigns. Believing the hype was almost the rational thing to do.
What no one anticipated, however, was the impact Campbell’s Law would have on the industry.
In his 1979 paper titled ‘Assessing the Impact of Planned Social Change’, social scientist Donald Campbell argued “the more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor.”
Campbell used the example of a city’s crime rate to show that a reduction in the overall number of crimes recorded doesn’t necessarily indicate an improved social environment but instead serves to a reflect the impact of procedural changes introduced to make the number look better. Police may, for instance, downgrade the seriousness of certain crimes or change the rules as to what should be reported.
The point Campbell makes is that when performance is measured solely from a quantitative perspective, it leads to gaming of the system. Participants find ways to accentuate the metric being measured, delivering what appears to be a good result on the surface.
Businesses disrupted by quants—or big data—are particularly prone to such gaming of the system. When metrics become an indicator of success, players in the game start finding ways to deliver the desired results without much consideration of the overall performance.
One game leads to another
As big data has disrupted advertising, we’ve seen various forms of gaming the system play out across the industry. The first and perhaps most rudimentary example would be the example of selling likes online. The moment entrepreneurial types realised that social media likes were a commodity some brands and online influencers were willing to pay for, packages offering likes popped up online. It didn’t take long for brands to cotton onto the fact that millions of likes were little more than a vanity metric, particularly useless when fake. Attempts to game the digital system certainly don’t end there.
Part of the reason why Google constantly updates its algorithms is to stay a step ahead of SEO consultants trying to take shortcuts to reach the top of Google rankings. Glen Maguire, the founder of SEO firm Clickthrough, says that one of the major challenges he faces when picking up a new client is unravelling all the dodgy backlinks bought by “lazy optimisers” looking for a quick buck from a digitally naïve business owner. While this approach works in the short-term, Google eventually cottons on and penalises the client’s website. And by that stage, the SEO company has likely moved on to its next mark, leaving the business owner with the mess.
However, buying links and likes remains only a minuscule part of the broader picture. Undoubtedly, the most successful and intricate form of gaming the system is found in the black market ad fraud industry.
Bots created by sophisticated criminal networks across the world mimic human behaviour and illicitly funnel billions of dollars of ad spend into their coffers every year. The ‘Bot Base Line’ report published by the Association of National Advertisers (ANA) in the United States recently found that online ad fraud is likely to cost advertisers an estimated US$6.5 billion in 2017. This is down from US$7.2 billion lost in 2016 but make no mistake, the fraudsters are only becoming more ingenious and today’s bots have become much better at mimicking human traffic. Much of the same technology that powers AI interfaces and digital algorithms are now also being used by criminals to make their bots look and smell a lot like human clickers. With this in mind, it’ll be interesting to see if that US$6.5 billion estimate isn’t perhaps adjusted at the end of the year, by which time investigators would’ve unravelled some of the more elaborate schemes syphoning away ad dollars right now.
The domino effect
Part of the reason the ad fraud syndicates have been so successful is due to the CPM (cost per 1,000 impressions) model upon which online advertising is often sold. Fraudsters have essentially taken advantage of the industry’s almost singular obsession with this metric and extracted illicit value out of it. But the CPM model hasn’t only inspired action from fraudsters; in fact, it has directly influenced the way business is done, websites are made and how marketers choose their partners.
GroupM global executive vice president of brand safety John Montgomery goes even further, saying in an interview with NZ Marketing that at least some of the transparency issues blighting digital advertising are attributable to “slavishly following a low-cost principle and measuring on CPMs”.
One place where this is always apparent is when marketers pitch their media business, he says.
“The agency pitch cycle has dropped from around five years to three years and a big reason for this is to maximise cost-effectiveness or cost efficiency,” he says.
“Whenever we have a new pitch, there’s a requirement to lower the CPMs and we then have to go to our media partners and negotiate very hard to bring the costs down. When those pitches happen in a category, like a big consumer group, and one advertiser does that, the other advertisers often follow in order to make sure their pricing is competitive as possible.”
Montgomery says this then has a knock-on effect in that publishers are forced to introduce new ad zones to make up the loss in advertising revenue. At best, this creates an annoying ad-heavy experience for the users or, at worst, it leads to ads being served on sections of a website never seen by human eyes.
To show how far the CPM game can be pushed, Montgomery and his team recently ran an experiment in which they constructed a faux campaign entirely out of invalid or unviewable ads.
“You can take your CPMs down to zero if you want to but don’t expect your ad to be seen,” he says.
When you extend this issue across commercial categories over a number of years, you end up with what has today manifested as a viewability crisis serious enough for the IAB New Zealand to dedicate an entire event to it in February.
The general consensus among the panellists was that it was important for all the players in digital marketing to accept the IAB’s oft-ridiculed standard that an ad can only be considered viewed if 50 percent of the pixels are viewable for a duration of one second in display or two seconds for video.
At the time, Moat partner manager Dave Goodfellow conceded the standard might be lax but added that marketers can always ask more of their providers.
“The important thing is to find a baseline that makes it possible for us all to work together,” Goodfellow said. “If there are people who want to go beyond that baseline and have a better quality audience, then they should be able to do that.”
The point here is that IAB’s standard is not an objective, but rather a minimum requirement to ensure all providers are playing by the same rules in the local market.
What makes the viewability issue so imperative is that the problem is rampant across both local and international sites. So pervasive is the problem that ANZ head of digital marketing David Gascoigne says he’s seen viewability rates as low as 40 percent on some of New Zealand’s biggest websites.
“Imagine paying TVNZ to air your TVCs and they come back to you, saying that they only managed to air 40 percent of them but they’re still charging you 100 percent – it makes no sense at all,” Gascoigne says.
Paying for peace of mind
Whenever there’s a mess, you’ll have someone willing to clean it—at a price, of course. And when it comes to the viewability debacle, advertisers have a number of cleaning agents at their disposal.
Moat, Integral Ad Science (IAS) and Comscore all offer tools that gauge viewability, giving advertisers a semblance of security that their ads are at least making it into the peripheral vision of a web user.
But this peace of mind comes with a price tag the marketer is required to pay. The services offered by these providers are added to the already-complex digital media supply chain and marketers can then choose whether or not they’re willing to pay the commission fees attached to having their impressions guaranteed as viewable.
What is somewhat unusual about this is that marketers are essentially footing the bill for a problem on the media owner side. And what further complicates the matter is that there’s still little in the way of third-party measurement when it comes to the online duopoly of Google and Facebook (and, increasingly, Snapchat).
In all other media channels (except perhaps outdoor), media owners pay for at least some independent reporting of their viewership or readership numbers, whereas the viewership and usage numbers for the big international sites generally come directly from those big international sites.
To put this into context, imagine the industry uproar if, for instance, the magazine or television industry suddenly told marketers that they would now be reporting their own numbers and that marketers would have to pay Comscore to ensure the ads were actually airing in the right place.
So why the double standard? Why are digital providers allowed to mark their own homework while other media owners chip in to pay the bill for independent analysis? Part of it is down to momentum. Marketers and agencies seem willing to forgo certainty to be on these digital channels simply because they are so popular. And you only have to look at recent growth in revenue to see this ‘crisis’ isn’t harming Google or Facebook (at least not yet). Also, it happens to be a little more difficult to measure digital audiences than anticipated.
“It’s tricky because there will always be variations, even from third-party suppliers,” says Gascoigne. “We see variances in Google Analytics and Adobe Analytics numbers even with the same code being applied to the same pages and actions across the site. [It’s] all down to the way in which these two vendors collect the data.”
The difficulty of the task aside, arguably the most obvious reason we don’t have independently provided numbers is that the likes of Facebook, Google and Snapchat have been resistant to allowing third-party verification of their audiences, preferring to keep their cards very close to their chests. And a few prominent voices across the industry, Montgomery’s among them, have raised the opinion that this breeds mistrust.
“When there isn’t third-party measurement allowed on a platform like Facebook, Snapchat and most of Google’s properties, there’s always a concern about what’s going on that we don’t know about,” says Montgomery.
That said, the digital juggernauts are starting to give the other players a few momentary peeks, with one example being Facebook’s commitment to an audit with the Media Ratings Council.
While a good first step, Montgomery doesn’t believe this is enough to ensure actual transparency in the industry.
“It’s after the case and it’s also not real auditing,” he says, explaining that a report released once a campaign has run does little to give marketers any security in the fast-moving digital space.
For this reason, Montgomery is a proponent of appending what he calls “a brand-safety tag” to ads running across social media networks.
“If we had a tag on those ads, then at least we would’ve known about it or we would have been complicit in it. Without any control, it’s an incredibly uncomfortable situation.”
A common argument against allowing a third- party to tag ads in Google, Facebook or Snapchat is that this could place the data of the subscribers in each of those channels at risk. While a leak or privacy breach would be extremely damaging to these social media giants, Montgomery believes it’s well within the capabilities of the companies to keep the information secure.
“Even if we were allowed to append a tag to an ad, there would be extremely strict agreements in terms of what would be allowed to be collected. Remember, this is not personal information… They’re not sharing names, addresses or any personal details. What they’re sharing is where the ad appears.”
Playing it safe
Knowing where ads appear is important not only to ensure viewability but also brand safety, the latter of which is particularly pertinent at a time when countless brands are being called up for advertising alongside objectionable material.
Interestingly, the introduction of a brand safety tag might not only be in the best interests of clients but also the media owners. As seen in April when myriad advertisers suspended advertising on YouTube and Google’s display network as it came to light that their ads were appearing alongside extremist and misogynistic content, brand safety is an area advertisers aren’t willing to negotiate on.
Since this scandal, YouTube has taken drastic steps, flagging large swathes of its online creators as ‘non-advertiser friendly’ in a bid to coax advertisers back to the platform.
Gascoigne says he understands why so many advertisers are unwilling to risk their reputations on the channel.
“If they don’t have the skillset and the understanding of how to manage transparency and brand safety, you’ll find a lot of businesses are turning and running, and that’s really the feeling I’ve had from the industry of late.”
He says ANZ is still using the channel but has taken a slightly more cautious approach.
“At the moment we stick with premium placement. We’ve got a 200-strong whitelist of top-performing and safe YouTube channels that we run a lot of our activity on.”
This still gives ANZ scale, but it doesn’t necessarily bode well for the channel as a whole.
Part of what makes YouTube so popular with younger consumers is that it offers edgier, racier content that might not be considered safe from a traditional branding perspective. And if these content creators are precluded from making money on account of strict enforcement of brand safety rules, then they’ll either have to find a new site that offers monetisation or stop producing their videos. Neither of these scenarios is ideal for YouTube—or for brands looking to engage with younger consumers.
Changing the conversation
Despite transparency and brand safety issues running rampant across the industry, marketers aren’t giving up on digital. Instead, they’re finding ways to measure the effectiveness of campaigns without paying attention to the dodgy metrics floating around.
The reason why we haven’t seen a mass exodus of advertisers from Facebook, despite its admission of a string of measurement errors, is that advertisers aren’t all that worried about basic view metrics as long as the campaign still delivers the results initially intended.
“With any digital media channel, you need to make sure the budget you are spending is delivering to the business objective,” says Gascoigne. “You need to take insight from metrics like views and engagement but judge the success of the activity in an action. For us, the importance of our cost-per-acquisition indicators are the number one priority, and yes we look at certain metrics with a ‘lens’, because at the end of the day you need to see the returns.”
In much the same way that online discussions have shifted from impressions to viewable impressions and then to safe viewable impressions, we’re starting to see measurement language evolve.
Rather than talking about reaching millions of people, marketers and programmatic providers are increasingly talking about return on investment and attribution models that claim to show the link between an ad and a consumer action.
The only question is how will these metrics be gamed and how many millions (if not billions) of dollars will marketers lose because of it?
Beware the new walled gardens
The lack of third-party verifcation offered by the big digital players is a concern. But an industry source, who agreed to speak to NZ Marketing under the condition of anonymity, believes marketers also need to be wary of the walled gardens being put in place at agencies.
The source says it’s become common for some of the bigger holding companies to set up independent programmatic offerings that they own and operate within the group.
In some instances this can become problematic, the source says, because you have what amounts to media arbitrage.
“You’ve got a media agency body buying space, sitting on it and selling it on to clients for a margin. Back in the day, this was called space farming and it was outlawed. This included things like agencies buying up stacks of billboards and onselling them. Space farming is not how we do business, and yet, the norm, at least in certain agency programmatic units, is that model.”
The counter-argument to this, of course, is that the agency is taking on all the risk by buying the inventory and should, by virtue of this, be allowed to make a return out of that. But the source rejects this reasoning because it essentially removes the objectivity of the media agency, which is meant to act in the best interests of the client.
“You’re not making an independent investment decision. You’re making a decision based on the inventory you’re trying to get rid of rather than what’s best for the client.”
The source says many clients are being caught up in these scenarios because they don’t fully understand how programmatic providers make money and therefore don’t know what questions to ask.
This is part of the reason he encourages marketers not to put pen to paper before asking the following four questions of an agency:
- What demand-side platforms (DSPs) are we actually using?
- Is there any financial benefit, direct or indirect, to the agency in using these DSPs?
- What’s the ownership of the DSP? Is there any shared ownership at a holding company level?
- If there are charges being added for technology and data, what specifically are you paying for?