You Can't Legit Compete Against Companies that Cheat

How do you make every single blog post "go viral?" Right, use bots. You can't rely on enough humans to like all your posts and share all of them out. How do you boost the number of streams of your new single to make it appear so popular it gets the attention of record labels? Right, use bots (like Travis Scott did ). How do you make more revenue and reliably hit growth rate targets promised in the hockey stick charts you showed in your investor deck? Right, use bots. Humans are not reliable, and you can't force them to come to your site, let alone get more and more of them to come to your site so you can hit growth targets, to get your bonus payments. But bots are very reliable. Bots are limitless, and also cheap. You can buy as much as you need to hit those growth targets. This is exactly what has happened, not just in digital advertising, but in other sectors too (hint: where Silicon Valley funded startups have been active).

"It's hard to build a business without cheating, because you're competing against venture funded companies built on cheating."

You can't compete against companies that cheat

Over the last 20 years, I have witnessed first-hand examples and heard stories relayed to me after the demise of legitimate startups. These startups failed because they tried to build their company honestly, without cheating. They simply could not compete against other companies that did cheat; digital metrics are the easiest to cheat. Venture funded companies are incentivized to cheat because there is no way to achieve the growth that Silicon Valley demands, just relying on humans to visit sites, stream music, buy more, etc. This twitter thread from Professor Bruce Clark from today perfectly illustrates the challenge of building a company based on realistic metrics. Professor Clark writes [there are simply] "too many streams pursuing the same number of eyeballs. Here's another piece from earlier in the year talking about the competition [The Economist, Feb 2022 - Disney, Netflix, Apple: is anyone winning the streaming wars?]"

These mainstream companies are not venture funded and are likely trying to build businesses without cheating. But they face the reality that there are finite humans, who spend finite time with media. And all of it is limited to 24 hours in a day (at least in this known universe). And the growth rates are small and slow, definitely not enough to support "hockey stick" growth charts shown in investor decks.

The chart above, based on Nielsen data, shows the trend of media consumption over the last four years (they don't have 2022 data yet). Two key things to notice: 1) the total time spent did grow, but slowly, from 11 hours 6 minutes a day on average in 2018 to 13 hours 21 minutes per day in 2021 - about a 20% increase; and 2) the dark green segment, representing "Internet Connected Device" grew most rapidly, rising more than 200% -- i.e. doubling -- in the same timeframe. This correlates to the increase in streaming that everyone intuitively knew was driven by a lot more people staying at home during the pandemic, which started in 2020.

Data from TVision (left side) shows that weekday streaming essentially doubled during the daytime hours. But the shape of the curve remains the same; in other words, most humans still stream the most in the evening hours. The InsiderIntelligence data (right side) shows the growth in CTV ad spending, rising from $6.4 billion in 2019 to $14.4 billion in 2021, and projected to hit $21.2 billion in 2022, according to the IAB . It may all seem plausible to you, especially if you are desperate to buy CTV ads.

Every ad fraud case in history is the same ... falsified metrics

How do I know? I compiled it here -- a Google spreadsheet "History of Ad Fraud" with links and examples dating back to 2008. Boris, Asher, Daniel, and Matt were openly selling traffic for websites. "If they [the buyers] wanted real humans, they [should] pay more and buy from other sites. They just want low cost and that's what I sold." Ad impressions sold by AppNexus were 92% fake, so obvious that even they could see it. After they purged their exchange from 260 billion impressions per month to 20 billion, in 2015, they gave no refunds to advertisers that bought the fake ads. Outcome Health faked the metrics about the number of screens they had in doctors' offices, falsified the performance, and also cooked the books. Ozy Media imploded in a week, after being outed for inflating every metric from website visits, to Facebook followers, to YouTube views. Venture back companies were incentivized to cheat in order to hit their growth projections. You could, and still can, buy everything you need -- from traffic, clicks, views, likes, follows, installs, and even sales. No those were not real sales, just falsified attribution to claim credit for causing the sale, when the sale had already occurred.

The result of twenty years of cheating ...

What happens when "everyone's doing it" and the cheating of digital metrics has been going on for 20 years? Well, everything today looks "normal" especially for the young marketers who started their careers within the last 10 years. All they've even known is super large quantities of ads to buy - like 10s of billions; super low CPM prices - like $1 to $3 CPMs; and high click rates on ads - like 1% - 10% CTRs. It seems entirely foreign to them if they saw real publishers with "only" millions of pageviews, CPM prices in the $30 range, and clicks in the 0.1% range. Their expectations of being able to buy billions of ads, at dollar CPMs, and get greater than 1% click rates are out of whack by at least an order of magnitude -- i.e. a factor of 10. Ad impression quantities should be in the billions, not 10s of trillions; CPM prices should be in the tens of dollars not ones of dollars, and click rates should be in the 1/10th of a percent range not 1% range. I know because I've been at this for 25 years. I've been doing digital marketing longer than some of these young marketers have been alive.

Let me end this article with some more data which you can use to gut-check things for yourself. The table below shows the top sites and apps in the U.S. using "last 30 day" numbers. Let me draw your eye to a few things. The blue numbers on the far left are the row numbers. At the very bottom you will see there are 107,052 total rows. I am only showing the top 75 and then it jumps to the bottom so you can see the totals. Those show that there's over 9 trillion bid requests observed from these sites and apps in a 30 day period, and 11 billion unique cookies. These are U.S. only numbers. Considering there are only 300 million people online (out of 350 million people total), that's about 37X (11 billion / 300 million) more cookies than persons. Again you can rationalize all these numbers yourself, with things like "there's more than one cookie per person," etc.

Another thing I will draw your eye to are the yellow highlights. These are what I consider "anchor points" and fairly reliable numbers. I look at legit publishers like cnn.com, nytimes.com, washingtonpost.com, espn.com and see their unique cookies per month numbers. That gives you a sense of the share of the 300 million online persons that these sites see every month. Everyone checks the weather, right? So weather.com sees 80 million unique cookies in a month, usatoday.com sees about 60 million unique cookies in the U.S. in a month, and so on. Let me know what you think. And let me know if you want to have a copy of the spreadsheet to play with the data yourself. And feel free to share this article out to others you think should see it.