What’s Next With Demographic Targeting?

Demographic targeting has long been one of the key tactics for advertisers when aiming to reach specific audiences and tailor their marketing strategies accordingly. However, with the evolving landscape of data privacy, the methods of demographic targeting are undergoing significant changes. In this article, we will explore what the future holds for this particular advertising technique.

Current methods of demographic targeting

Deterministic targeting in closed and open ecosystems

Demographic targeting can be implemented in a deterministic way through closed ecosystems, where users are required to provide basic demographic information during the registration process. Examples of closed ecosystems include social media platforms and membership-based websites. This tactic can also be used in a limited way on the open web, by using cross-site user IDs, typically cookies, to extrapolate demographic information collected on one website to other contexts. 

In the simplest example, a data management platform ties the demographic data of a given user to their ID in its database. Then, it can provide user ID lists with this additional data to demand side platforms (DSPs) for activation and targeting on the web. This activation works in such a way that the DSP receives bid requests from its inventory partners which may include IDs of users from the lists, and when they do, the DSP can bid appropriately.

Probabilistic targeting based on content analysis

Probabilistic targeting involves predicting user demographics based on the content a user engages with. By analyzing the types of articles, videos, or products users interact with, advertisers can make educated guesses about their age, sex, interests, and other demographic factors. This approach eliminates the requirement for explicit user input and relies on data patterns and algorithms.

There are a few options on how to use this technology in practice. One example is using content categorization by third-parties, such as supply side platforms (SSPs). These companies have direct relationships with publishers. It allows them to analyze the contents of publisher websites and select relevant URLs to activate via Deal IDs passed to buyers. Such an analysis can be improved by the deterministic data of some publishers which can be provided to their partners. This approach is never fully accurate and relies on a buyers’ trust in the company doing the categorization.

Alternatively, DSPs can build web scraping, contextual analysis engines themselves, and use them for predicting demographics of audiences behind a given URL. This approach is even less precise from a demographics perspective. Typically there’s a limited, or no direct relationship between DSPs and publishers, making it more difficult to reinforce classification models. 

The outlook for the cookieless future

The shift towards data privacy and increased user control is expected to limit the possibilities of applying demographic targeting across the open web. Cookies are small files that allow websites to remember information about the user. Later on, this information can be used in the first-party context (first-party cookies) or in the third-party context on external websites (third-party cookies). Tracking individual users across different websites using cookies will not be possible in Google Chrome after 2024. So what are the options for demographic targeting in the cookieless environment?

Email-based identification with questionable scale

Deterministic demographic targeting will still be available in a fairly unchanged manner within  closed, so-called “walled gardens” ecosystems. If a major social media platform is able to collect its users’ demographic data, it can store it within a user profile which uses an email address for identification. Later on, it can be used to inform personalization within the very same platform.

However, even if demographic data is collected within one portal, advertisers will face challenges when trying to link this data to a universal, cross-site ID and use it across the internet. Just from a technical perspective, in order to make it work, the user has to log in to each and every website, where data tied to this ID is intended to be used, with the very same email address. This is not something that users want to do―in most cases they want to access both publisher and advertiser content without the need of providing personal data. Even if they are forced to do so, they are reluctant to provide the same email address on each website.

There’s also an elephant in the room regarding the privacy aspects of building a cross-site user profile linked by email address―arguably it is an even more invasive mechanism than third-party cookies have ever been. Similarly, there are multiple initiatives run by browser vendors that aim at limiting this form of tracking. Most notably, hide-my-email from Apple that allows users to generate anonymized email addresses protecting them from tracking, or FedCM from Google providing tools supporting federated login use cases without leaking data to trackers.

A content-based approach with limited precision

Seemingly, the content-based approach is not highly dependent on cookies. Provision of demographic information by users in the first-party context will still be available, making it possible for SSPs to categorize URLs by characteristics and sharing dedicated deal IDs to buyers.

In practice, though, these models will be less accurate, as the amount of data available to reinforce categorization models will be limited to first-party only, instead of application of third-party data sets. On the other hand, from the scale perspective, this still seems the best approach moving forward for demographic targeting, despite its limitations.

The reduction of precision of probabilistic classification of content will also affect the DSP-based approach. They will lose access to any data allowing them to reinforce their models, since the possibility of directly using first-party data from publishers is already limited.

Demographic targeting might get deprioritized

It is clear that the scale and accuracy of all demographic targeting approaches will be limited. Many years of using this targeting tactic will result in great efforts towards its preservation among advertising vendors. However, technical limitations imposed by browsers and regulatory authorities, along with limited user willingness to share their personal information, may result in its relegated importance in the ecosystem. 

Alternative technologies, such as contextual targeting, behavioral targeting, and retargeting may be more frequently chosen as viable alternatives, depending on the use case. While the landscape may become more challenging, innovative solutions will continue to emerge to ensure the effectiveness and relevance of advertising campaigns in reaching the desired target audiences.

‍