The Downfall of Cookies

In the previous article, A Glimpse of AdTech, we skimmed over cookies and their decline. Today, we’ll dive deeper into why cookies have been facing challenges in recent years and explore potential alternatives for maintaining relevant ad targeting.

Overview

Cookies

Before we start, it’s important to understand the two types of cookies:

First-Party Cookies. Set by the website the visitor is currently browsing. They are typically used for basic website functionalities such as session storage.
Third-Party Cookies: Set by websites the visitor has previously browsed. These cookies are used to track visitor behaviour across multiple websites.

In this article, we will focus on cookies used for tracking visitor behaviour, referred to as tracking cookies.

While tracking cookies have been useful for improving ad targeting, they present several problems:

Privacy. Tracking cookies were often set without explicit visitor consent, which is invasive as it can reveal sensitive information about interests.
Transparency. Visitors often had little to no visibility or control over which companies were tracking them and how their data is being used.

Regulations

As privacy concerns have increased, regulations such as GDPR and CCPA emerged.

In practice, these laws mean that cookies (both first-party and third-party) must now comply with requirements such as:

Visitor Consent. Websites must obtain explicit consent from visitors before placing cookies on their devices, except for cookies that are strictly necessary for the functioning of the website
Transparency. Websites must provide detailed information about the types of cookies they use, their purpose, and how visitor data will be used.
And more besides

So every website owner concerned by those regulations has to comply. Which usually translates into a pop up when landing on the website:

There’s one key player we haven’t mentioned yet: the browser.

Browsers

While they’re not directly responsible for compliance, browsers play a significant role in facilitating website compliance by providing features like:

Cookie Management. Modern browsers offer visitors options to block, delete, and manage cookies, including third-party cookies used for tracking.
Private Browsing Modes. These modes do not store cookies, history, or other personal data beyond the current session.

Moreover, most major browsers have started to phase out third-party cookies. For example, Firefox blocks third-party cookies by default, meaning many tracking cookies are automatically blocked unless visitors explicitly allow them.

NB: While the main deprecation efforts are around the third-party cookies, even first-party cookies that are not used for basic website functionalities now needs explicit visitor consent due to GDPR and other regulations.

Conclusion: Due to increased awareness of privacy issues, evolving regulations, and actions taken by major browsers, the use of tracking (first and third-party) cookies is facing significant challenges.

Alternatives

Remember the equation of AdTech

Better Targeting = Better Ads = More Money

So while cookies decline, we need to find way to maintain our ad targeting capabilities.

There are several options out there, let’s talk about some of them.

Contextual Targeting

With contextual targeting, the ads are chosen based on the content of the current page the visitor is browsing.

This means the ad is tailored to what the visitor is seeing right now, not what they have seen on other websites. In other words, it’s based on current content rather than past behaviour.

It works by:

Analysing the content of the page
Categorising the page’s content with relevant keywords
Matching the ad inventory of the website to ad campaigns with similar keywords

NB: The first two bullet points rely heavily on machine learning algorithms. For example, image recognition algorithms are used if the page displays images

Pros	Cons
Privacy-Friendly: It doesn’t rely on user data or past behaviour Relevance: Ads are more relevant to the content the user is currently viewing.	Less Personalization: While ads are relevant to the content, they may not be relevant to the visitor’s interests Content Limitations: If the algorithm misinterprets the content, irrelevant ads might be displayed.

So, while contextual targeting is purely content-based there are other initiatives that use behavioural analysis as the cookies do.

Federated Learning of Cohorts (FLoC)

As part of the Google Sandbox initiatives, FLoC is a module implemented directly within the browser.

Over a rolling time window, FLoC assigns the visitor to cohorts based on their browsing history.

In practice, it uses machine learning to cluster visitors with similar behaviour patterns. The browser then shares the cohort ID with the website, which forwards it to the ad network/advertisers/…

NB: The cohorts are recalculated regularly as the visitor’s browsing patterns change.

But FLoC raised concerns:

Transparency. Visitors had limited visibility and control over which cohort they’re placed into.
Privacy. It was potentially possible to trace the cohort ID back to individuals by combining it with other information, such as browser characteristics.

Because of those issues, Google decided to give FLoC up and developed the Topics API as an improvement.

Topics API

Also part of the Google Sandbox initiatives, the Topics API is a browser-based API that websites can use to retrieve information about the general interests of their visitors, based on theirs recent browsing history.

Each week, the browser selects a few topics based on the visitor’s browsing behaviour from the recent websites visited. These topics are stored and can be accessed by websites.

But we’re still clustering visitors based on their browsing history. How is it different from FLoC ?

There are 2 main reasons:

Broad categorisations. The list of topics is maintained by Google. They aim to provide enough granularity for effective advertising while ensuring visitor privacy by avoiding overly specific categories. Thus, a visitor might be assigned to “Traveler” or “Martial Artist” rather than a specific cohort based on detailed browsing behaviour.
Amount of data shared. The browser stores a visitor’s top three topics (one for each of the past three weeks) and shares a randomly selected one with websites when serving ads. This means that advertisers only get a snapshot of a visitor’s general interests, rather than a detailed profile of their browsing habits.

Benchmarks

Note that we lack benchmarks to assess the performance of these alternatives compared to third-party cookies. There are likely two explanations for this:

Simply because the alternatives are still fresh.
Because these benchmarks need to be audience-agnostic and not all the audiences react the same way to ads. For example, some audiences might be highly responsive to targeted ads. This makes it much harder to assess the performance of an targeting method.

Conclusion

These alternatives highlight potential trends that could shape the future of AdTech. That is to say:

A refocus on first-party data (Contextual Targeting, First-Party cookies, …)
A more “privacy-friendly” approach to fetch third-party data (FLoC, Topics API, …)

Which one do you think will prevail?

Hope you enjoy it, reach out to me if you’d like to dive into a specific topic!

https://www.linkedin.com/in/malo-le-goff/

If you want to receive additional tech insights, subscribe to my newsletter!