July 22, 2021

How marketers can achieve scale and precision in a world without third-party cookies

Share on

“One of the biggest challenges for the open web going forward is bringing the fragmented ID space under a single, standardized umbrella and convincing the ad tech market to adopt it.”  

Sarah Polli

Director, MarTech Solutions Hearts & Science

The future of data modeling

With third-party cookies finally crumbling, four kinds of data will hold the key to targeting, measurement and attribution going forward: first party (1PD), second party (2PD), third party (3PD) and contextual.

To find potential consumers and reach existing ones, advertisers will need to be flexible about which datasets to leverage for specific use cases and how to combine them.

Pillars of the consumer data ecosystem, post-cookie deprecation
First Party 1PD

This information that advertisers collect directly from consumers, which can be leveraged to provide the most personalized ads and content. It encompasses data on anonymized users’ behavior, actions and interests, which is generated via usage of websites and apps and typically collected through site analytics. It also encompasses data on authenticated (signed-in) users, which is typically more valuable due to its accuracy and persistence. 1PD can also be collected through loyalty cards, surveys, call centers and mailing lists. In a cookie-less world, 1PD will be critical for seeding look-alike models to enable targeting at scale.

While 1PD is the most accurate and least expensive type of data to leverage for paid media, CPG brands and other advertisers may lack access to it. That’s because vending machines don’t require an email address or phone number in order to dispense a Pepsi, while Pepsi-selling retailers, such as Walmart and CVS, collect the customer data directly and do not share with the CPG brands.

Second Party 2PD

This data from another company that collects it directly from users and sells it to advertisers and is also referred to as “platform data.” It’s largely used by advertisers to enrich their 1PD for deeper insights and increased scale. The largest 2PD sources are Google, Facebook and Amazon, as well as publishers like The New York Times and Condé Nast. Retailers like Walmart and Target are also entering the space and providing valuable customer data to CPGs and other brands with limited 1PD.

Third Party 3PD

This is obtained from companies that aggregate customer data from other sources (e.g., public records, registration data or other proprietary sources, such as credit cards) and is typically licensed by advertisers for use. Third-party data cannot be traced back to an individual user. It can be 3P cookie-based (for now), mobile ID-based or non-cookie-based and shed light on purchase behavior, offline transactions, location history and more.

While 3PD provides the most scale, marketers have limited visibility into data collection practices and quality. Additionally, advertisers pay a CPM for usage of this data for activation (typically done directly in a DSP), which can be costly.

Contextual targeting

This is a form of targeted advertising based on webpage content. The type of content being consumed is the relevant signal, not the identity of specific visitors. Many brands employ contextual targeting to avoid having their ads shown alongside undesirable content. (This practice is sometimes referred to as negative contextual targeting.)

Since it doesn’t rely on cookies or run afoul of privacy regulations, contextual targeting is getting considerable attention right now, though the basic approach is as old as internet advertising. For two decades, advertisers have been providing a keyword list to publishers, which is matched against keywords in specific webpages; this signaled where ads could be published. (A deodorant brand might have wanted to show up in articles that mentioned “gyms,” “exercise” or “sweat,” for example.)

But since the match had to be exact for impressions to be delivered, publishers couldn’t fully monetize their content.But with recent advances in machine learning and usage of natural language processing (NLP), contextual tools have a more sophisticated understanding of context and sentiment. Partners that offer this technology use a variety of techniques, including page crawlers and tags, to extract text and digest it for classification. The bottom line is that an article doesn’t have to explicitly mention “gyms” for technology to infer it’s of interest to fitness enthusiasts.

Real world scenarios

Marketers will need to use these data types in tandem to be successful in a post-cookie world. Here are three scenarios for how it works in practice:

1PD and 2PD

A telecommunications brand sends 1PD segments to DV360, Google’s digital buying platform, and applies Google’s affinity audiences to the campaigns to increase scale.

2PD and contextual

A CPG brand with no customer data of its own targets prospects on Facebook, in addition to using contextual segments through The Trade Desk.

1PD, 2PD, 3PD and contextual

A fintech brand has lots of 1PD, but when it develops more specific segments, the scale is significantly reduced. Using Forbes’s 2PD, the brand can increase its segment size. Meanwhile, to bring new users to its site, it also uses contextual segments in The Trade Desk and collects a 1P cookie on new visitors to increase its 1PD. Finally, it partners with IRI for transactional data to upsell existing customers.

What steps should marketers take to prepare for the future?
Test contextual targeting

In the absence of a firm date for third-party cookie deprecation, advertisers should start testing contextual targeting today, whether that’s alongside their 1PD or not. There are many vendors in the ecosystem that specialize in contextual, such as Peer39 and Semasio, and it’s also offered by all the main ad verification partners.

Review data collection strategies

It’s also important to review your data collection strategies to ensure that data from your owned properties is consented to and compliant with relevant privacy regulations. From there, brands should conduct regular audits to ensure that customer data is stored and maintained securely. When reviewing your data collection strategies, also think about how you can leverage more signals (e.g., on-site actions from authenticated users and content consumption rates) to feed into look-alike models.

Audit your data usage

Next, audit your current usage of data to understand how your current data partners leverage 3P cookies. For any that rely excessively on cookies, examine the feasibility of shifting to 1PD or 2PD sources to replace them.

Upgrade to Google Analytics 4 (GA4)

If you currently use Google Analytics, develop a plan with your Google team on upgrading to GA4 by the end of the year. (Hearts can assist with this transition, so reach out to your account lead to learn more.) Also stay tuned for testing opportunities with industry solutions such as the open-web identifier from IAB Project Reach and The Trade Desk, Unified ID2.0, which would map to your 1PD, and Google’s interest-based group targeting (FLoC) via DV360, Google’s digital buying platform.

Develop testing frameworks

Finally, work with your Hearts team to develop testing frameworks to determine the best mix of data for a variety of use cases. Each campaign may require a different mix to meet its goal.

In making these shifts, it’s important to be open, flexible and—above all—optimistic.

The demise of third-party cookies is a historic change for the industry, but we should view it as an opportunity to lean into machine learning, NLP and other technologies that can help us build scalable audiences with much higher accuracy.

Written by

Hearts Marketing Science Team
Stay connected.

Sign up for our newsletter and stay up to date with our latest industry news and content.

Become a MaxBounty Affiliate Today

Click here to sign up