Most marketers had been happily paying for programmatic advertising for the last decade, very proud of themselves for being “digitally transformed.” They were also happily using vanity metrics like CPM prices, number of impressions, and click through rates because those were easy to measure and easy to report. Buying digital ads became as easy as playing a video game, with colorful dashboards that showed them what great discounts they got (“cost efficiency”), the number of impressions they bought (“reach”), and how many clicks they got (“performance”). But this triple cocktail of low price, large reach, and high performance was so addictive because every part of it was faked by fraudsters.
The low CPM prices were only possible from fraudulent or fake sites that plagiarized all their content or used no content at all. Real publishers with real human audiences had real costs of producing the content; so they could not sell ads for very low CPM prices. Further, there’s a finite number of humans that visit their sites every month; so they could not magically manifest a lot more reach. But fake sites could easily do this by buying traffic and doing audience extension. No one can force a herd of humans to all go to the same site at the same time to increase its traffic and audience; but it takes no more than one command line to instruct a vast botnet to generate a large number of pageviews on a site — exactly the amount that was paid for. And these same bots click on the ads too. Not too much or else that would be suspicious. Bots tune their click through rates to be in the 5 – 15% range, which is always higher than real human click rates. This way, marketers are tricked into thinking ads on fake sites are performing so much better than ads on real sites with real humans, so they allocate more or all of their budget to programmatic channels, which are teaming with such fake and fraudulent sites.
Do you see how this all worked together? Larger quantities of ad impressions, lower CPM prices, and better performance — indeed the illusions of vast reach, cost efficiency, and performance — led to what is now known as “digital marketing’s lost decade.” When “programmatic” ad buying really took off in 2012-13, the disparity from reality really took off as well. Note the green and yellow lines in the chart below — those represent humans’ usage of the Internet, social media, and mobile. Those two lines are pretty much flat across since 2012-13; indicating that real humans’ usage had all plateaued, already maxed out. But the blue line representing digital ad spending continued upward. How can this dissociation from reality be explained? Easily, with bots. Bots are simple software programs that can be remotely controlled to automate browsing (load more pages) and simulate desirable human actions, like clicks on ads. It was technically trivial to simulate all the things that marketers wanted to buy — more reach, more clicks, lower prices.
The rise of programmatic also corresponded with marketers’ digital transformation. Marketers were shifting more and more budgets from offline channels into digital. Conveniently the vanity metrics in digital reinforced the “correctness” of those decisions to “go digital” so they were reported up the line and became the default metrics for reporting on the success of digital marketing campaigns. In the process, marketers lost sight of the real metrics that should matter — business outcomes. Clearly outcomes were not as easy to report — after all, marketers would have to figure out the complex attribution models that determined which sale was driven by which ad or action. What they forgot is that advertisers don’t actually need to know “who” bought the shampoo — which individual person; advertisers really only need to know that people who were exposed to ads bought shampoo at a higher rate than those who were not shown the ads. This was a perfect example of “too much data” from digital channels leading marketers down a rabbit hole that yielded worse insights.
Some marketers have had the courage to run “turn off” experiments with their digital media. What was interestingly consistent is that all of them found that turning off their digital ad spending didn’t change business outcomes — eBay (2015), P&G (2018), Chase (2017), Uber (2019), AirBnB (2020). So what were they spending millions of dollars on in digital, if it were not producing real, measurable business outcomes? We may never know. But what is clear is that more marketers need to check their own digital spending more closely, and do things differently than they have been doing for the last decade — or shall I say “lost decade?”
MORE FOR YOU
Marketers should pay higher CPMs by buying ads from real publishers with real human audiences. You know you have to show your ad to a human before you can get any kind of business outcome right? Showing ads to bots, no matter how low the CPM prices, will drive no incremental business for you.
Paying higher CPM prices don’t necessarily mean greater costs. That’s because CPMs are unit pricing (cost per thousand digital impressions). If you bought fewer ad impressions, even at higher CPMs, your total cost could actually be lower. You don’t need the vast quantities or enormous “reach.” It’s not real reach, it’s just the illusion of reach, if you’re not “reaching’ humans anyway. You don’t need to buy as many ad impressions to reach real humans. Humans tend to visit a small handful of mainstream sites repeatedly. Even though they do visit long tail sites for niche content once in a while, the “at-scale” quantities of impressions from the programmatic long tail are also an illusion, that conveniently helped fraudsters feast on marketers’ ad dollars for the last decade.
Finally, accept lower click through rates. Humans click on ads very rarely (when was the last time you deliberately clicked on any ad?). But the lack of clicks does not mean the campaign performed poorly. But the presence of clicks faked by bots does mean the campaign performed poorly. Those clicks are not real, and the high CTRs (click through rates) don’t mean real performance. If you understand the above, you will also understand that the single most important factor in digital marketing is getting your ad in front of a human in the first place. Everything else — like targeting, viewability, click rates, etc. — is secondary. Smart marketers are ditching the crappy ad tech targeting (that costs extra) and simply showing ads to Safari and Firefox users because they are humans. They are also getting a great deal — 50-70% lower CPMs — because other marketers are not bidding on Safari and Firefox users due to the lack of targeting. But showing ads to humans always beats targeting for business outcomes, because the data used for targeting may not be accurate and you may be targeting bots pretending to be certain audience segments.
After the last decade of digital transformation, marketers should now pull themselves out the “lost decade” of doing digital marketing based on vanity metrics – low prices, vast reach, high clicks. Time to think differently and do different digital marketing. Pay high CPM prices (ads on real publishers) with finite reach (real human audiences) and low clicks; showing ads to humans is the first step towards better digital marketing, indeed marketing that actually drives real business outcomes.