What’s next with behavioral targeting?

Brands with a high level of awareness are often perceived as more credible than their competitors and can see the knock-on effects of increased customer loyalty, market share, and sales. Staying top of mind is a top priority for most businesses. If customers are interested in a certain product, you need your brand to be among the first ones they consider—and it’s no good waiting for the moment when they are ready to buy.

Recent research indicates that 59% of customers prefer to buy from brands they already know.

The effectiveness of any branding campaign depends on the correct definition of the target group. A proper understanding of your target persona makes it easier to choose effective forms of advertising, as well as the right messaging and placement. This is particularly true for retargeting campaigns but is also important for branding campaigns. 

One pitfall is that we often base our human understanding on assumptions that are influenced by biases or misconceptions. Considering demographic, socioeconomic, and psychographic features is useful to some extent, but the latest marketing strategies more frequently focus on analyzing online behavior. User personas are then built from real-life and irrefutable actions undertaken by users, rather than from our potentially biased opinion of who they are.

How does behavioral targeting currently work?

Behavioral targeting is a technique used by advertisers and publishers to present users with tailored advertisements and marketing messages that align with their online browsing habits. In essence, this type of targeting is predominantly based on data associated with a user's behavior on the website. Trackable events like pages visited, previous search queries, duration of time spent on a website, clicked ads, and website visits are crucial information that enables user segmentation. 

Users are clustered by dedicated ad tech vendors into segments defined by behavioral patterns (for example, people who travel a lot, people who like biking, or people who often return to the same product category).

There are a variety of audience providers on the market, each offering numerous segments using their own taxonomy. In fact, one vendor can offer over 10,000 different audience segments! Such a variety allows advertisers to choose which segments they believe are most suitable for their campaign.

However, this diversity also causes some confusion because it is often not known how segments are constructed. It is also not always clear how a segment from Vendor A is really different from a segment from Vendor B, because they might have similar names. In response to this issue, there has been an effort from IAB to standardize the taxonomy of audience segments, including behavioral ones.

‍

“IAB Audience Taxonomy 1.1”

‍

When talking about reaching the target audience provided by third-party vendors, each platform and channel has its own characteristics. In essence, the vendor has to provide a list of the users to be targeted. There are different methods for how this list might be constructed and delivered.

• In the web ecosystem, the vendor typically provides a list of cookies
• Device IDs, such as AAID or IDFA, are used in mobile devices
• CTV uses a combination of device IDs and third-party IDs.

Let’s take a closer look at how data flow works in the web environment from a marketer’s perspective.

Imagine that you are a sports clothing company and you are launching a new line of running shoes. After reviewing the list of available audience segments, you decide that the “jogging lovers” category will be the most suitable persona for your ads. You then get a list of cookies that you need to target.

This process raises some data privacy issues. Are users (the actual people behind each cookie) conscious that they will be packed in the bundle and sold to third parties? Do they know that data collected on a given website will be mixed with data coming from other websites? How did they agree to participate in this process? Is there a reliable way for users to opt-out?

Such data-privacy issues are why traditional behavioral targeting methods are reaching their expiration date.

The phase-out of cookies and its implications

The phasing out of third-party cookies has already caused great debate among audience providers. The majority of them rely on third-party cookies or fingerprinting methods, but these solutions might be pushed out of the market due to increasing pressure from publishers, users, regulators, and other parties. 

This leaves marketers all over the world with a pressing question: Will contextual targeting be the only option left on the table in branding advertising? 

As data privacy is clearly one of the most important advantages of contextual targeting, we need to bear in mind some of the disadvantages. For example, as a standalone method of targeting, it only works at a specific point in time—at the moment when the user is reading a particular article. Currently, it is possible to tie this information to the user’s cookie and use it to build a behavioral profile. However, with cookies gone, it will not be possible at scale.

Are there other options for cookieless behavioral targeting?

Thanks to joint industry efforts, there are three crucial, privacy-friendly technologies on the horizon:

Topics API. Each website opted-in for this API will be mapped to a specific, general domain—like ''sports'' or ''fashion''—from the ~470 available in the current taxonomy. Based on visits to these websites, users will be assigned with the five most relevant topics for each of the last three weeks, which can then be retrieved by ad tech vendors. In order not to increase the fingerprinting surface, ad tech vendor can only receive one topic from each of the three weeks. This topic will be randomly selected from those assigned to the user, or, in 5% of cases, it will be a totally random one.  You can learn more about Topics in this article.

Protected Audience API. In short, this API allows the ad tech ecosystem to conduct audience-based auctions without relying on third-party cookies. Based on granular analysis of user behavior on a single website, they can be added to custom, ad tech-defined ''interest groups'' which are accessible for targeting by this vendor on the internet. This API supports both upper-funnel (behavioral) and lower-funnel (retargeting) use cases. The former can be achieved by using data from a given publisher's website. The latter takes data from the advertiser website. Importantly, thanks to the deprecation of third-party cookies and limiting alternative individual user cross-site tracking technologies, only data from a given website can be used for assigning users to groups. Additionally, groups cannot be mixed between different websites. This helps to keep the transparency of data ownership and usage. You can learn more about the Protected Audience API in this article.

Seller-Defined Audiences (SDA). Two previous examples were from the Privacy Sandbox from Google Chrome, but there are other options, such as Seller-Defined Audiences which is championed by IAB Tech Lab. This concept allows publishers to categorize users on their websites according to either standard IAB taxonomy or custom taxonomy of their choice, and to pass this information to the demand side. Importantly, it does not require any user-level ID, meaning data leakage can be minimized. SDA supports contextual taxonomy—allowing publishers to signalize current user browsing context—as well as audience taxonomy for more in-depth behavioral user analysis. You can learn more about SDA here.

Behavioral targeting is not over, but it will change

Without cookies, the possibility of detailed, cross-site user data collection will be significantly limited for any third-party vendor—especially on an individual level. What remains rock-solid is first-party data, which, for behavioral use cases, is collected directly on publishers’ websites. This data will be the holy grail of the cookieless world.

Combining new technologies like Topics API and PA API with publisher input seems to be an attractive alternative to cookie- and ID-based solutions. The publisher provides in-depth information about the user behavior on the website, stating areas of interest for a particular user. These signals can be enriched with broader context from Topics API, which leverages what the user is doing outside of the publisher’s website. Combining these two approaches and using Protected Audience API to build relevant target segments looks like a promising solution for behavioral targeting in the cookieless future.

‍

Sources:

1. Why is brand awareness important?, 16 February 2022
2. IAB Audience Taxonomy 1.1, 30 March 2020

‍