900,000 Americans Tell Google They Want to Try, Buy, or Quit Heroin Every Year

I use ad data to try to save their lives

Photo: Tobias Schawrz/AFP/Getty Images

A number of the kids I graduated high school with are dead, in jail, or in rehab from heroin.

I’m friends on Facebook with those who stopped using. They post about sober anniversaries, featuring pics of Narcotics Anonymous coins in upturned palms. And I notice when they stop posting. It means they’re no longer incentivized to share the current version of themselves, which usually means they’re back to using.

Many relapsed addicts go dark on social media, and turn to a place where they can share things they don’t want anyone to know: Google.

Every year, approximately 900,000 Americans tell Google they want to use, try, buy, or quit heroin.

I know this because I serve ads on Google to them. I buy keywords I imagine they might type into the search box. When my keywords match with their searches, my ad shows up as their first result on Google. When they click the ad, their ad click data — which includes the exact words they type into Google — passes into my system.

Here are some word-for-word searches addicts Googled before clicking the ads:

I run five different ads to five different groups:

  • People who tell Google they’re going to try heroin for the first time (including searchers looking to prepare or cook heroin, and those wanting to make the leap from pills to intravenous use).
  • People who ask Google where they can buy heroin (including people traveling to new cities where they have no dealer connections).
  • People asking Google about the current percentages of fentanyl in local heroin (Fentanyl is a deadly synthetic opioid, and it is fatal in small doses).
  • People asking Google how heroin will make them feel.
  • Other current users (including those looking to stop using, get help through withdrawals, or troubleshoot a collapsed vein or abscess).

Each group sees a different ad. The ad clicker is directed to my website, which contains a phone number for a substance abuse hotline. From there, the data passes to Google, and Google passes it to me.

This data includes the words clickers type, but also their web browsing history, where they work, the size of the company they work for, whether they rent or own a home, whether they’re married or single, whether they have kids, how old their kids are, the apps they use, whether they’ve recently gone through a “major life event,” whether they drive a Chevy (or prefer Ford), any degrees they hold, whether or not they graduated high school, and more.

Ad click data provides nearly two decades worth of the world’s confessions, held in indelible ledgers by a corporation.

I am having a series of one-sided conversations with the people I know who became addicts, and so many people like them. I learn more about them than I ever could from what they post on social media, and I try to use these insights to help them.

Ad click data is in aggregate. Google doesn’t provide marketers like me with your personal identifiers. This means that even with all of the data we get when you click, Google will never provide us with your name, email, or phone number.

But there’s an easy way around this: through a call-tracking software that anyone buying Google Ads can use. It’s called Callrail and costs only $45 a month.

By syncing Google Ads with Callrail, your name and phone number is tied to your click (and all your other data) when you conduct a certain action on a website that an ad directs to. (Clicking a phone number on the website, for example).

Callrail provides marketers the data that Google would never dare to. Yet Callrail still withholds your exact location from us.

But there’s also an easy way around this: through reverse searching the name or phone number.

Here’s one scenario that someone who is not me might use to abuse the system:

A heroin dealer builds a simple website: a single page that contains a phone number and the words “Substance Abuse Hotline” under it.

The dealer signs up for Callrail and buys a pool of 8 forwarding numbers. Now, when an addict calls the number on the website, their call gets routed to the dealer’s phone without the caller knowing.

To drive addicts to his website, the dealer runs a Google ad. Let’s say he targets his ad to Long Islanders who live in a 10-mile radius around his house. He buys keywords like “where to buy heroin in Long Island.” Now when addicts Google this, the dealer’s ad shows up as the first result.

What ad click data reveals is a boon to addressing human pathologies. Yet most people who need this data don’t know where to look for it, how to interpret it, or that it even exists.

The dealer writes an ad that promises the addict will get help by clicking it and calling the number on his website. This is a lie, but the dealer is confident that if someone is desperate enough to Google “where to buy heroin in Long Island,” they won’t hang up once they realize they’re on the phone with a dealer instead of a substance abuse hotline.

By running the Google ad for long enough, the dealer will eventually know every prospective customer within a 10-mile radius.

By following these steps, the dealer avoids whatever lax hurdles are put in place by Callrail and Google to prevent this kind of thing:

  • Google allows you to buy keywords like “where to buy heroin,” but doesn’t allow ads that say anything about heroin. (The dealer’s ad copy promises salvation. He doesn’t use the word “heroin” in his ad).
  • Google vets the destination phone number their internal forwarding numbers direct to. (Callrail doesn’t).
  • Google doesn’t allow any copy on a website ad that suggests drug selling. (The dealer doesn’t use copy besides “Substance Abuse Hotline”).

I understand that this kind of click data can be abused by those with bad intentions. I’ve written about this before and still don’t think it should be overlooked. The potential of ad click data to understand large-scale trends in an unprecedented way — to be utilized by policy and decision makers fighting the opioid epidemic — outweighs the risks of it being abused.

Ad click data provides nearly two decades worth of the world’s confessions, held in indelible ledgers by a corporation. In terms of importance, there is no dataset in history that comes close. It’s so powerful that the clicks on marketers’ ads bought with it make up 90% of Google’s revenue, which came in at more than $136 billion last year.

Yet what have we been doing all this time with these incomparable insights about humanity? Selling people stuff. This is a big mistake.

Ad click data is a microscope. It allows the viewer a way to see society’s maladies in a way that is completely hidden to epidemiologists and psychologists and politicians and economists.

But as long as Google’s ad system has been around, the microscope has been pointing at the wrong things. If marketers seem perfectly willing and capable to use ad click data to their own ends, why can’t doctors and teachers and governments use it to help people who are suffering; people who need help but don’t know where to turn to find it except Google?

Like the microscope, what ad click data reveals is a boon to addressing human pathologies. Yet most people who need this data don’t know where to look for it, how to interpret it, or that it even exists.

In nearly every one of my experiments targeting heroin users in the United States, mine was the only ad running.* We need to take a look at the opioid epidemic through the microscope of click data, because the traditional methods of gathering data on heroin users are flawed.

Government and nonprofits currently gather data on heroin users in the United States from:

  • Estates
  • Medical data
  • Crime stats/police data

These only provide insight into heroin users who’ve died or been arrested. To understand users who are alive and haven’t been arrested, researchers use survey and poll data. But this data is flawed because people lie on surveys and in polls — especially about something like illegal drug use. It’s the same reason my high school classmates disappear from Facebook when they relapse. Because they want to avoid their audience.

Google searchers believe there is no one seeing what they type. Because of this, they are more honest with Google than with anyone in their lives, any other online platform, and any survey or poll.

Click data can give us a heat map of U.S. cities most in need of a safe injection site, for example. Or the demographic segments most in need of education about fentanyl. Or the income brackets most at risk of taking the leap from pills to shooting up in every city in the United States. Or the websites people who are at risk of relapsing browse the most. Or the unmet needs of those facing withdrawals.

Here are two rhetorical questions:

How can we expect to end the opioid epidemic using flawed data? Why aren’t we using the most accurate and robust dataset (Google’s ad click data)?

And three more:

Why is almost no one running ads to addicts? Why should someone searching Google for shoes, or directions, or a degree, be shown 20 different helpful ads, when someone telling Google they’re going to use heroin is shown none?

If ad click data — the world’s thoughts, amassed — is so powerful that it’s used to sell people nearly any product or service, shouldn’t it also be used to sell life to someone in the market for death?

*Nonprofits who receive Google Grants can advertise on Google’s dime. There are some recipients of Google Grants who advertise to addicts, nationally. I don’t count them, because Grant recipients face strict restrictions from Google that limit their data gathering capabilities — their data is limited and flawed.

Founder of a NY search ad agency (like we need another). Finding humor in marketing tech’s depravity. AdTech writing @ NY Times. patrickberlin08@gmail.com

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store