Failure to Click: Why the Metrics of Online Advertising are Broken
The online advertising market has never been bigger. It’s already the single biggest advertising medium in many countries (including the UK and Australia) and it is likely to be bigger than TV by 2020. And yet, the way we currently measure the success of that advertising is deeply flawed.
The vast majority of online advertising campaigns are measured using two metrics: how many times an advert was loaded on a web page (page impressions), and how many times it was clicked. Of these two, the click is by far the most popular measure of success.
Google has built a $400bn empire, 90% of it on the back of getting people to click on adverts. There’s an intuitive sense to this. When someone clicks on a link or a banner, they’re obviously signalling their interest, right? Except that, more often than not, they aren’t. Security researchers estimate that at least 25% of the clicks on adverts are fraudulent. Hundreds of unscrupulous publishers pay black market “click farmers”, usually based in developing countries, to visit their low quality web sites (often consisting entirely of content scraped from other sites) and click on the ads.
These click farmers run a new kind of sweatshop. Hundreds of desperate people are packed into unventilated rooms full of computers. Their mission – to click on thousands of ads in return for $1 or less per day. The click farmers sell these clicks at small premium to the publishers, who then make a larger premium from the advertiser. But since these clicks are fraudulent, the advertiser is essentially paying for nothing. Some poor guy in a Bangladesh slum does not want to buy an American insurance policy, even though he clicked on an ad for one.
Ok, so how about we forget the clicks and just measure how many people have seen the ad? Alas, the prognosis for page impressions is even worse. When executives at Heineken dug into their online ad spend, they discovered that only 20% of the ads they paid to flight were actually seen by people. The Alleged $7.5 Billion Fraud in Online Advertising How Much Of Your Audience Is Fake? Some of these missing ads were simply outside of the area of the screen being viewed by the reader, but the vast majority were “viewed” only by bots.
These synthetic web users can be programmed to visit websites by the thousand. They are just sophisticated enough to fool advertising systems into displaying adverts to them. As with click farms, the mechanism is simple but effective: dodgy website owners buy traffic on the cheap and sell it to advertisers at a fat margin.
The technology may be new, but the concept is as old as time: a shell game in which the advertiser is the unsuspecting mark. We could look to technology for the solution. After all, it caused this problem. Except that this is almost identical to two other plagues of technology – the computer virus and the spam email.
No matter how hard the forces for good fight, there are always going to be more bad guys, and they are always going to find new loopholes. This is an arms race which cannot be won by technology alone. We must look beyond the limited metrics of page impressions and clicks.
These are relics of the 1990s that do not serve us well in the 21st century. By continuing to use them as the standard by which we measure success, we play into the hands of the fraudsters. None of this is to say that clicks are worthless. Search engine marketing remains one of the most powerful advertising channels ever created. But we cannot treat all advertising as we do search advertising without the consequences we are currently facing. What will replace the click and the impression? It’s not at all clear. Those two metrics are so simple and well understood and their replacements are unlikely to be as intuitive and easily grasped. But as an industry matures, so must its tools. One idea is to tie brand advertising to incentivised web surveys. After each burst of advertising on a site, visitors are polled about brand recognition and results are fed back to the advertiser. Sites receive a performance bonus if recognition is high.
Another strategy is to focus on affiliate and partnership programmes with trusted publishers. Bots can create traffic, but they will never buy a product or visit a physical store. If we can tie a real person to a physical action via a digital channel, then the bots don’t stand a chance.
Reward programmes are another good mechanism. By rewarding people directly for genuine interactions with a brand, and tying them back to physical purchases and activities you create a feedback loop that bots can never fake. These mechanisms are a lot harder to get right, and even harder to measure accurately. They assume a level of partnership between publishers, agencies and advertisers that is currently rare to nonexistent.
But the myth of easy advertising success is just that – a myth. We can continue to turn a blind eye, and happily peddle clicks and impressions to oblivious clients, but one day there will be a reckoning. Three years ago Kellogg pulled its entire ad spend in-house after discovering rampant fraud by publishers. Kellogg will not be the last. It’s time for us to ditch the click before it sinks our industry.