Note: all materials are reproduced under Fair Dealing

In order to understand my own perspective on targeted ads, I wanted to see how targeted ads perceive me.

About a month ago, I deleted almost 10 years of ad data on my personal Meta account because I was becoming increasingly uncomfortable with my information being processed in the United States, especially considering the precarious state of Canada-US political and economic relations. I’m also pretty conscious about denying permissions for ad tracking across websites and apps.

All of this, however, does not necessarily mean that ads don’t appear at all or that my preferences are being respected.

The ads

In the past month since deleting my ad data, my Meta account data states that I have interacted with (clicked on) 34 ads on my Facebook account. 54 companies have used my Meta accounts (one personal Facebook, one personal Instagram, and one professional Instagram) to influence their targeted ad campaigns.

Here are the 34 ads I’ve apparently been clicking on while I’ve been on Facebook over the past month. You can right click an ad and open it in a new tab to read it fully:

These are the 54 companies that have used my Facebook information to inform their ads over the past 3 years:

  • Tangerine
  • AirBnB
  • Porter Airlines
  • Playnow.com
  • Canva
  • Mark Carney (campaign for Liberal Party leader)
  • The Crocodile Seattle
  • Provident Entertainment
  • Bricknell’s Men’s Products
  • Sony Music Entertainment
  • RightNow Media
  • Pattison Food Group
  • Elevation Church
  • BMO Financial Group
  • Amazon.com
  • Coach
  • GrubHub
  • Jackbox Games
  • Canadian Tire
  • Envision Financial
  • Virgin Plus
  • KOHO
  • Wise
  • Williams Sonoma
  • Scene+
  • Square
  • Coast Capital Savings
  • Live Nation
  • Staples
  • Wealthsimple
  • Uber
  • Grab
  • Neo Financial
  • Old Navy
  • ASOS
  • Groupon
  • Remitbee
  • Linktree
  • GoodRx
  • Meta for Business
  • LinkedIn
  • Ancestry.com
  • Equitable Bank
  • Callia Flowers
  • Premier Productions
  • Remitly
  • Etsy
  • RIU Hotels & Resorts
  • Le Choix du Président
  • DoorDash
  • Showcase
  • Drive with Lyft
  • Ticketmaster
  • Cineplex

The challenge

We have 88 datapoints of who I am based on my normal interactions with advertising content on a service I use every single day.

Since I already know myself very well, and because targeted ads are based on patterns and algorithms that aren’t usually accessible publicly, I’m going to provide these datapoints to the mightiest algorithm I have available: ChatGPT.

I’m going to ask it to guess the following 10 pieces of demographic information about me that many advertisers use to inform their ad campaigns:

  • Age
  • Gender
  • Location
  • Occupation
  • 3 to 5 hobbies/interests
  • Income range
  • Ethnicity
  • Political affiliation
  • Marital status
  • Religion
  • Any other inferences on my demographics based on the datapoints?

The challenges I anticipate are that the 34 ads were not targeted to me based on any of these demographics, these were just the ones I happened to interact with, and the list of the 54 companies are just company names and not actually ads themselves so they aren’t based on ad content.

Before you proceed!

Take a look at some of, if not all of the dataset. Try to jot down a list and compare it to what ChatGPT suggests I am and what the truths are!

The findings

The algorithm has spoken, and here is what it guessed:

  • Age: 25-40
  • Gender: Male
  • Location: Greater Vancouver
  • Occupation: professional OR student in finance, education, or technology
  • Hobbies and interests:
    • Finance and investing
    • Education and personal development
    • Travel
    • Tech & Creativity
    • Community & Social Causes
  • Income: $50,000-$80,000
  • Ethnicity: visible minority
  • Political affiliation: Liberal or progressive-leaning
  • Marital status: single or in a relationship, no kids
  • Religion: Possibly Christian or faith-inclined, but open-minded

Surprisingly, it got only three of its guesses wrong:

  • Age: I’m actually 22
  • Hobbies and interests: I’m not interested in finance, especially not in investing
  • Income: I make $20,000 to $30,000 a year (since I’m just a student)

You can read the full analysis ChatGPT made, including the justifications for each of its answers here.

What this means

Even though ChatGPT got almost everything correct, that does not mean that ChatGPT is the algorithm that Meta uses for its advertisement strategies.

It does suggest that if I allow it to be the case, I am a lot more predictable than I think, and algorithms are processors of patterns about ourselves that we miss in all of our humanness.

Based on these patterns, we can even make more specific predictions like how I’m not only a Christian, but I’m an open-minded Christian towards other religions. It’s fun to make a demographic profile and all, but these even more targeted answers can reveal stuff about ourselves that we might not be comfortable with or things we might not even be aware of.

This is a lot of information and accuracy for a limited amount of data. This is only advertising data, not data on the user-generated content I interact with or any information that I’ve given up about myself. How much more specific of a profile can we get if we consider my entire online footprint holistically?

In what ways can these patterns be helpful? Or for a malicious purpose?