The advertising funnel, probability, and you

Why do I keep getting these “Nigerian millionaire” emails? And why do I keep getting “phishing” emails purportedly from my bank? How did millions of Britons fall for categorically false messages on Brexit, and how did millions more Americans abandon their lifelong principles to vote for Trump?

This is a story about a very old marketing concept, called the advertising funnel, sometimes called the purchasing funnel. It is also a story about some basic probability math, and the reliability of human folly.

I have written in recent past posts how political advertising unavoidably affects you and I, first through our collective delusion that is doesn’t work when it actually does work. Next, I wrote a series of posts describing one way to mathematically predict how people change their minds on consumer and political preferences. In this post, I will explore this old marketing concept of the “funnel” and how it was used most recently to suck people into both the British vote to exit the European Union and in the rollover of long-held Republican principles to swing the 2016 primaries, followed by the general election, to Donald Trump.

The advertising (or purchasing) funnel

Around the turn of the 20th century, a pioneering advertising executive named E. St. Elmo Lewis began to outline his principles of consumer behavior, which evolved over the years into many different iterations of a “funnel” concept, where a product or idea is “sold” by first getting a prospecting message to as many people as possible, and then winnowing down that audience through various “filters” until you get to the much smaller number of customers who will “bite” on the final purchase, or alternatively adopt a new idea being marketed. One version of this funnel, using the language of Lewis, looks like this:

Advertising funnel

The concept is simple. You cast as wide a net as possible to get people into the top of your funnel through basic awareness of your product or idea. Then you winnow this large group down into the smaller group of those who show some interest, assess their realistic desire to “own” the product or idea (a yet-smaller subset), and then finally get an even smaller subset to take the action to “buy.”

That Nigerian email

Depending on the demographic and market served, each “filter” of this funnel has a probability factor that predicts how many people will “fall through” to the next filtering stage, until you finally get to the bottom. In the classic “Nigerian Millionaire” scam, billions of emails have been sent out to the “top of the funnel” so far. Most of these messages get filtered out early by credulous readers, but a small percentage get through each layer of filtering, through successive emails that tap into interest and desire. The math for the Nigerian scam is similar to the math of lotteries, which I have explored in earlier posts. If you start with one billion emails, and 99.999% of them are filtered out at some stage of the funnel, this still leaves 10,000 people who get “sold” on the scam, which is enough of a success rate to keep the emails coming. An even smaller percentage of success would likely keep the scam alive.

Good “phishing” scams, which trick you into giving up bank, credit card, or email passwords through deceptive emails and websites, have been confirmed to have delivered access to the Democratic National Committee server emails to Wikileaks and Russian operatives closely allied with the Kremlin during the 2016 election season (but, interestingly, apparently not to the personal Clinton family server that received so much congressional attention). These scams have a much higher rate of success than the Nigerian millionaire trick. I am personally very cautious about this type of scam, but I once received an email “from my bank” that was so well constructed that I had to make the kind of technical analysis of hidden content that most people do not know how to implement in order to figure out the nasty intent of the email.

But legitimate “advertising funnel” marketing is used on all of us, all of the time. How many times have you seen the same television advertisement for a particular insurance product or a prescription drug with a long list of scary side-effects? These advertisements keep running because they are at the “top of the funnel” which eventually draws in enough paying customers to more than cover the expense of the advertisements. Much of our television, radio and internet economy is financed by “top of the funnel” prospecting costs.

And I admit that I am personally not immune from a well-played, straightforward advertising funnel. While visiting a Subaru automobile dealership a few years ago to try out my own well-thought-out plan to achieve a “deal,” the sales representative threw me off-guard with an out-of-the-blue comment: “I’ll bet you are a teacher,” he said. In retrospect, I can objectively see how the Subaru marketing strategy was very effective in implementing these “funnel” steps to get to my demographic, and then to me in a series of steps.

We are naive if we think that we are immune to effective marketing. I have written extensively in this blog on the concept of human volition (or “choice” or “free will”) and how it is not as “free” as we have usually convinced ourselves. [1] The same holds true for political advertising and its end-effect of getting us to vote in particular ways. Are you capable of reviewing the ways in which you have been “played” over the years, to good or bad ends?

The Brexit and Trump Funnels

In the case of the lead-up to the Brexit and Trump elections, we can now establish with a high degree of certainty that a significant “advertising funnel” started with some 50 million stolen Facebook profiles. [2] These data profiles passed through a web of internet marketing companies and their funders that “coincidentally” included some politically-active U.S. billionaires (e.g., the Mercer family), the internet data firm Cambridge Analytica, the alt-right Breitbart political website, noted Trump campaign adviser Steve Bannon, a British Brexit campaigner with a Russian wife and deep Russian financial interests (Arron Banks) and several Russian oligarchs whose names keep popping up. This is not “conspiracy thinking” here; it is basic funnel marketing.

The funnel for both the Brexit and Trump campaigns started with psychologically-tailored messages to at least those 50 million Facebook users, which expanded into at least 126 million Americans alone receiving messages tailored to their specific demographic. [3] In addition there were likely millions more Twitter and email accounts specifically targeted, working their way down in probabilistic filters from “awareness” to “interest” to “desire” to “action.” The last filter was designed not just to get some people to vote FOR the Brexit or Trump position, but also to encourage part of the “funneled” to just not vote.

Note that we also need to add in some of known ways to “get to” political message consumers that make the message more effective on a particular demographic than a simple appeal to reason. Appeals to the darker side of people’s bigotries and fears usually work well, and both the Brexit and U.S. presidential elections saw a lot of that. Basic misinformation that could easily be disproved with very simple research also works extremely well here, because most people don’t research or read outside of the “comfort zone.” Both of these tactics were widely implemented by the Brexit and Trump campaigns.

Add to this the apparent knowledge in the U.S. election of end-of-campaign Clinton state-targeted strategies retrieved from stolen emails of campaign strategists. This information is perfect “filter” programming information, and even conservative Republican observers with an understanding of Russian tactics and basic math have concluded that these efforts likely turned the American election toward Trump. [4]

The question is whether you think the probabilistic math of the filter in either the Brexit or Trump elections was more like the Nigerian Millionaire scam, with a very tiny chance of success, or more like the sophisticated phishing scam, with a much higher probability of success, or more like whatever it was that persuaded you to eat at the last new restaurant you frequented. It would only take a two percent “funnel conversion rate” (or even less in some U.S. states) to have altered the outcome of either election.

And unless you are under that “collective delusion” that political advertising does not work, then the probability that this specific Russia-Facebook campaign “worked” to affect two critical world elections is very high indeed.


  1. This post on choosing between a piece of chocolate cake or a piece of fruit is a good place to start. You can than follow the “Dice” thread icon at the bottom of each post or click on the “Volition” category to see all of the relevant posts.
  2. Graham-Harrison, Emma, and Carole Cadwalladr. “Revealed: 50 Million Facebook Profiles Harvested for Cambridge Analytica in Major Data Breach.” The Guardian, 17 Mar. 2018.
  3. Isaac, Mike, and Daisuke Wakabayashi. “Russian Influence Reached 126 Million Through Facebook Alone.” The New York Times, 30 Oct. 2017.
  4. Boot, Max. “Without the Russians, Trump Wouldn’t Have Won.” The Washington Post, 24 July 2018.

1 thought on “The advertising funnel, probability, and you

  1. Pingback: Penny-sucking economics – part 2 – When God Plays Dice

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.