How User Tracking Devalues Ads

Facebook recently took out full-page ads in the New York Times, the Washington Post, and the Wall Street Journal attacking the way Apple protects its users’ privacy. In the ad, they make the point that Apple harms Facebook’s ability to track people who see Facebook’s ads, and run personalized ads, which, according to Facebook, harms the effectiveness and thus the value of these ads.

This is kind of strange, if you think about it.

Why would Facebook take out a huge non-personalized ad to make the point that, for ads to really work, they need to be personalized? Why advertise in a newspaper if they think that personalized ads are so much more effective?

It’s because the idea that personalization increases the value of ads is wrong. Personalization harms the value of ads, because it measures the value of ads based on a metric that doesn’t really apply to most ads.

Personalized ads that use user tracking measure ads based on a direct causal relationship between users seeing an ad, and users acting on that ad by buying the product advertised in the ad. By that metric, the vast majority of ads just don’t work. People don’t see an ad for a product, and then buy that product immediately, or perhaps a few days later.

(In fact, every time scientists try to measure the effectiveness of advertising, it turns out to not be very effective at all.)

Instead, the way ads work is that when people decide to buy a product, they will have more trust in products whose ads they see consistently, and whose products they associate with publications they trust. In other words, if you consistently see a car brand advertised in the New York Times, you will assume that this brand is trustworthy. When you decide to buy a new car, you will have a preference for that brand.

This doesn’t just work for large publications and huge brands. If you see LTT consistently have sponsorships from Seasonic, you will be more likely to trust a Seasonic power supply for your next PC. If you see Kandji regularly sponsor Daring Fireball, you’ll remember their name if you ever need the kind of product they offer.

But you will never see an ad for Mercedes on a website, and then just arbitrarily decide to buy a Mercedes based on having seen that ad. You will never see a Seasonic sponsorship, and just randomly decide to throw out your old power supply, and buy one from Seasonic. You’ll never see a Kandji sponsorship, and just decide that you suddenly, urgently need their product. Thus, by the metric we value user-tracked ads, most of them have no value at all.

If Facebook wanted to increase the value of its ads, they would join Apple in fighting against user tracking, because in the end, it will increase the value of its ads. The less advertisers know about the direct causal effects their ads have, the higher they will value them.

If you require a short url to link to this article, please use

designed for use cover

But wait, there's more!

Want to read more like this? Buy my book's second edition! Designed for Use: Create Usable Interfaces for Applications and the Web is now available DRM-free directly from The Pragmatic Programmers. Or you can get it on Amazon, where it's also available in Chinese and Japanese.