Skip to main content

How Advertising Targeting Really Works

Every targeted ad you see is the output of a data pipeline that started long before the ad was made. Understanding it changes how you move online.


The Bid That Happened While You Blinked

Arjun visited a travel site. He searched for flights to Bangkok, browsed two hotels, and closed the tab.

He did not book anything.

A diagram showing data flowing from a person's device through ad exchanges to multiple platforms.

Within 200 milliseconds - faster than a blink - the following happened:

A tracking pixel sent Arjun's device ID, IP address, and the pages he visited to an ad exchange. His profile was matched against records held by three data brokers. An algorithm scored him as a "high-intent travel buyer, 25-34, mid-income." Eight advertisers bid in a real-time auction for his next page impression. The winner paid $0.003 to show him a hotel ad.

He saw the ad before the page had finished loading.

For the next two weeks, Bangkok hotels followed him across every app he opened.


What Is Actually Happening

200ms

The time it takes for a real-time ad auction to complete after you load a webpage.

Profile matched. Bid placed. Ad delivered. Before you finish reading the headline.

Source: Internet Advertising Bureau, Programmatic Advertising Guide, 2024
Data Signals

5,000+ Data Points Per Person

Major data brokers hold an average of 5,000 data points per individual - including purchase history, location patterns, browsing behaviour, health interests, and inferred political views.

Source: Federal Trade Commission Data Broker Report, 2024
Retargeting

70% Higher Conversion

Retargeted ads convert 70% more effectively than standard display ads. Products that "follow" you are not a coincidence - they are a deliberate high-ROI commercial strategy.

Source: Criteo Retargeting Performance Report, 2024
Lookalike Audiences

Your Data Targets Strangers

Platforms let advertisers upload a customer list and find millions of similar users. Your behavioural profile can be used to target people you have never met, without your knowledge or consent.

Source: Meta Ads Help Center; documented practice across all major platforms
Inferred Attributes

Beliefs Inferred Without Asking

Ad platforms infer political views, health conditions, pregnancy, and relationship status from browsing patterns alone. No survey. No disclosure. Just algorithmic inference from what you click.

Source: ProPublica, "Facebook's Secret Ad Categories", 2024

The Pipeline in Four Steps

Step 1: Collection

Every click, scroll, pause, and purchase generates a signal. Tracking pixels, cookies, app SDKs, and loyalty cards capture behaviour across thousands of sites and services. You do not need to be logged in - device fingerprinting identifies you anyway.

Step 2: Profile Construction

Data brokers aggregate signals from hundreds of sources into a single profile. They assign segment labels: "new parent," "financially distressed," "health-anxious," "politically right-leaning." These segments are for sale.

Step 3: Auction

When you visit any ad-supported page, your profile enters a real-time bidding auction. Dozens of advertisers bid within milliseconds for the right to show you one ad. The auction completes before the page finishes loading.

Step 4: The Ad

The winning ad appears. It looks like a relevant recommendation. It is the output of a multi-layer commercial system built around your behaviour.


Build Your Ad Profile

Answer 10 quick questions about your recent online behaviour to see the inferred profile that gets bought and sold on your behalf.


What That Just Showed You

Your browsing behaviour is the input. Your inferred identity is the product.

A few clicks is enough to assign you to segments covering income, health, politics, and emotional state. None of these inferences require your confirmation - they are sold as-is to advertisers who bid for your attention.

The same data profile is used for commercial ads, political messaging, and insurance pricing. What segment you fall into affects not just what you see, but what you pay and what opportunities are offered to you.


How to See Your Own Profile

You can view partial versions of your ad profile directly:

  • Google: myaccount.google.com/data-and-privacy - Ad Center shows your interest categories
  • Meta: Settings > Ads > Ad Preferences - shows your interest segments
  • TikTok: Settings > Privacy > Ads Preferences
  • Apple: Settings > Privacy & Security > Apple Advertising

You cannot see what data brokers hold. You can request deletion from some via opt-out directories, but it is not comprehensive.


Three Things Worth Doing

1. Use a browser with tracking protection. Firefox with uBlock Origin or Brave blocks most third-party tracking pixels. This reduces the signals entering your profile without requiring you to change your behaviour.

2. Check your ad settings on platforms you use. Reviewing what Google and Meta believe about you takes under five minutes and often reveals inferences that are surprisingly accurate - or surprisingly wrong.

3. Use incognito mode for sensitive searches. Health symptoms, financial problems, relationship issues - searching these while logged into a platform links the search to your identity profile. Incognito plus a private DNS reduces this.


One Question Before You Continue

Knowledge Check

A data broker labels you as 'financially distressed' based on browsing patterns. You never told any platform this. What is the most significant risk of this inference?