I’ve said over the years that programmatic advertising is an illicit drug spreading like cocaine through marketing communities everywhere for the last decade. Like crack cocaine it is hyper profitable for the dealers that sell it and ultra addictive for the users that buy it. Every user knows it’s bad for them and will kill them eventually, but they are so physically addicted to it that if they try to stop buying it they will have uncontrollable convulsions. I was talking about cheap ads sold through programmatic channels, not cocaine — programmatic ads are so addictive because every aspect of it gives marketers the high they’ve been seeking — enormous reach, low prices, and high engagement.
But the enormous “reach” is not from “ads shown to humans.” That’s because not many humans visit millions of long tail sites that they’ve never heard of and humans’ usage of the Internet, social media, and mobile devices have long since plateaued — i.e.maxed out — since 2011-12. The bot activity that makes up 2/3 to 3/4 of Internet traffic accounts for the massive quantities of ad impressions being transacted through programmatic channels. There are simply too many millions of sites and apps running ads through programmatic exchanges that no one checks. Automated checks are easily defeated by the fraudsters; and the botnets they use are tuned to bypass detection by the current fraud verification vendors.
So what’s a marketer to do? How does a marketer kick this nasty addiction? I understand and empathize with your predicament — the number of impressions, the low CPM prices, and the number of clicks have all been reported up the line for many quarters. If we reduce the use of fraud fueled programmatic ads, the number of impressions will go down, the CPM prices will go up, and clicks will go down. Right. That will all happen as you go through withdrawal. What follows is a three step plan that marketers have used to transition away from fake digital marketing back to real marketing.
Monitor continuously and remove fraudulent domains and apps yourself
Do you have fraud detection in place right now? Have they been helpful in reducing fraud? How do you know? I know your answers will be yes, yes, and uhhh. Virtually all marketers spending money in programmatic know how crappy it is, and have therefore spent extra money buying fraud verification, hoping to detect your way out of trouble. However, current fraud detection can’t catch anything , is prone to error , and still doesn’t help you ; they give you a percentage IVT after the campaign, so the only recourse you have is to try to get a refund for that percentage. Do they tell you which domains and apps were causing the fraud and how much? Do they explain why something is marked as fraudulent or not fraudulent? Ever wonder why their IVT percentages are always in the single digits? Is it because fraud is actually that low or simply because they can’t detect most of it?
Instead of the “black box” fraud detection vendors above, it’s better to have your own analytics for continuous monitoring. Analytics shows you the details so you can understand why something is fraudulent or not, troubleshoot problems as they come up, and see which domains and apps are cheating so you can add them to block lists, or remove them from include lists. This way, you can progressively clean your own campaigns while they are still running, rather than wait for post-campaign IVT reports that just tell you a percentage with no explanations or supporting details. Remember the Uber lawsuit against 100 mobile exchanges for outright fraud. Even if Uber eventually wins the lawsuit, most of those fraudsters’ companies don’t exist any more — they made off with the money and Uber will never get their money back. Better to clean the campaign while it is still running so less money flows to fraudulent domains and apps, and more money flows to better domains and apps.
Most people can’t quit cocaine, or smoking for that matter, “cold turkey.” You can’t stop buying programmatic ads overnight. But you can progressively clean your campaigns and make them better. As you go through this process, the number of ad impressions you buy will go down, the average CPM prices you pay will go up, and the number of clicks will go down too. This is because you are reducing the fake ads shown to fake users on fake sites that use bots to generate fake clicks. You are thus rectifying your analytics to be more realistic. Note that higher CPM prices are not a bad thing, when you are buying ads on real publishers with real human audiences. It also doesn’t mean higher cost because if you buy fewer ad impressions, even at higher prices, your overall costs go down. And you are wasting less money on ads shown to bots in the process.
Change KPIs to use reliable outcome metrics
The next step in kicking the habit of bad digital marketing is to update your KPIs to focus on outcomes. This seems painfully obvious. But too many marketers have relied on, and still report on, easy to measure quantity metrics like numbers of impressions and number of clicks (see above survey results from the ANA). They lost sight of real measures of business outcomes. Even though marketers know “conversions” are the most important KPI, the most used KPIs are vanity metrics — CPM (”cost per thousand”), CPC (”cost-per-click”), reach, and site visits. Using these metrics incentivizes bad behavior (buying more quantity, at lower prices, from programmatic channels) instead of real publishers with real human audiences.
Some marketers also say that because they don’t sell anything online, it’s hard for them to tie digital ads to outcomes — e.g. selling shampoo in physical stores. Media agencies do this all the time too. They send clients monthly excel spreadsheets that show the number of impressions purchased, the average CPM prices, and the number of clicks they got. None of these have to do with business outcomes, especially if you realize that ad fraud is responsible for the large quantities of ads, low prices, and high clicks.
Marketers that don’t sell directly online can still select metrics that are a good proxy for outcomes. For example, a medical device manufacturer counted on patients going into offline clinics that used their specific type of Lasik. It was hard to actually tie offline visits to their online marketing activities. So they chose a proxy metric — did the user click on the ad, do a zip code search on the site, and click on the listing of their local Lasik clinic? The two-step metric (zip code search and click to the local clinic website) was a reliable outcome metric because bots had no incentive to do those two steps. So bot activity was not messing up that outcome metric. That’s why it could be used as a reliable proxy for outcomes they would expect to see in the physical world. Each marketer should identify real outcome metrics or reliable proxies for outcomes to use to judge the effectiveness of their digital marketing programs, especially their programmatic campaigns.
You may find that getting far fewer clicks could mean you get far higher actual outcomes; those previous clicks were from bots anyway. And bots don’t convert. You may find that paying far higher CPMs, but buying far fewer ad impressions, still drives more outcomes, and lower overall costs. You won’t know until you update your KPIs to focus on real business outcomes, instead of the vanity and quantity metrics that were easy to report.
Optimize for real business outcomes
Finally, once you have the right KPIs in place, you should optimize your digital marketing activities to maximize your outcomes. Again, painfully obvious. But many many marketers doing programmatic advertising had been optimizing for “more engagement” — i.e more clicks — and “more cost efficiency” — i.e. lower CPM prices. What they were unintentionally doing was sending more money to bot-fueled sites that had higher click through rates (bots love to click ads, humans don’t). They were also allocating more budget to fake and fraudulent sites because those are the only ones that can afford to sell ads at very low CPM prices. Real publishers that have real journalists and editors making real content for real humans cannot sell ads for very low CPM prices. So paying higher CPMs on average by buying from better sites means more of your ads are shown to humans. That’s the only way you will get more business outcomes. Obviously, everyone knows ads shown to bots won’t drive more business, even though they drive nice large numbers in monthly excel spreadsheets.
The question is whether you are doing digital marketing or paying for big numbers in spreadsheets. The more you optimize for outcomes, the less you will buy from programmatic channels. Those drive enormous reach, low prices, and clicks, but they don’t drive better business outcomes for advertisers. You will see a drop in reach, and increase in CPM prices, and a decrease in clicks; but you will see a rise in business activity and outcomes. Kicking the habit of bad digital marketing is hard. And it will take time. But if you follow the three step process above, you will wean yourself from this addiction and get back to good health and real marketing.
I am optimistic that you can do it.