Cookieless Future: Good Inventory Hygiene, Privacy-Safe Targeting, and New Measurement Metrics

It was the industry-defining topic of the last few years: The digital advertising market is in a state of flux, with various changes taking place that will ultimately mean that the widely used third-party cookies will probably no longer be available beginning mid-2023:

• Regulatory changes, such as mandatory user consent for the use of cookies for marketing purposes (cookie banner)
• Technical changes, such as the launch of various browser initiatives to block third-party cookies
• Social changes, such as increased sensitivity to privacy in the digital space.

All these developments mean that digital marketing use cases based on third-party cookies will no longer be possible in their current form, or only to a limited extent. Remarketing, interest targeting, or frequency capping can be mentioned in this context. There is no way around looking for alternative solutions and approaches.

At the moment, the most important and promising initiatives to ensure targeted advertising in the future are Unique IDs, Google’s Privacy Sandbox, contextual targeting, frequency and conversion modeling, and conversion APIs. We are actively monitoring and evaluating these initiatives and innovations so we can assess which technologies work well—and which do not.

However, we are not only observing, but also working on our own approaches to react to the change processes for our customers in the best way and at an early stage. These aspects are relevant for this change:

• Less user and conversion data will be available.
• Relevant, privacy-safe data that can be used without restrictions is necessary.
• We will rely on new measurement and optimization metrics, such as the time on screen of display ads or audibility of video ads.

Based on this, we propose, in the future, we should:

• rely more on the use of privacy-safe data
• make it usable for automation
• improve the quality of our campaigns on the levels of inventory, targeting, and media buying.

The quality of the advertising inventory—i.e. the selection of all environments in which ads are published—is a central factor of every campaign because that’s where a campaign, and thus the advertising brand, is visible. That’s why, for us, high-quality inventory forms the basis for successful campaigns.

For this reason, we have the advertising inventory purchased for a campaign checked for performance values and scanned automatically and continuously via a script developed specifically for this purpose. If certain publisher domains do not meet our quality standards in terms of viewability or click-through rate, they are automatically flagged and added to the exclusion lists if necessary. This also applies to ad collisions, i.e. multiple appearances of the same campaign on the same page simultaneously. We record these collisions with our own technology based on the Google Ads Data Hub.

These measures have a major impact on inventory hygiene, as internal analyses show. A constant and automated check of inventory hygiene achieves noticeable changes in the purchased inventory: In the period we measured over several months, we blocked over 11 percent of the offered inventory due to poor quality using the above measures. This data used for quality control is all privacy-safe.

First-party data from premium publishers

In the targeting area, we see great potential in the increased use of publisher data—i.e. data collected directly by the operators of the websites on which the ads are played. In contrast to third-party data, this data is still available, is of a higher quality, and shows very good performance values. This case study with Miele explains how we, working with TX Group as a publisher, achieved an interaction rate four times higher than that of third-party audiences.

For the effective use of this data, close cooperation with the publishing partners is essential for us from the agency’s viewpoint: Since all publishers operate largely independently of each other and define the criteria of their audiences individually, there can be major differences. For example, the “New Home Seeker” audience at TX Group comprises different user groups than the audience of the same name in the Ringier network. To increase comparability between the networks, we held discussions with all Swiss premium publishers and thus standardized the individually defined target groups. Finally, with our in-house developed deal management platform, we bundle the deals with different publishers and address users automatically via the various networks.

Targeting based on hard data

Another promising alternative to third-party data at the targeting level is the use of hard and exogenous data such as geo-location, time, or weather, combined with dynamic ad playout. Dynamic messages gain significant relevance through location- or time-dependent delivery and can thus deliver better results. Appropriately developed concepts and infrastructures for maximum automated campaign delivery help to place the best-performing message in the right place at the right time. CSS’s multiple award-winning fall campaign has impressively demonstrated these functionalities.

The third aspect to our approach is to use newly measurable metrics to focus media buying optimization even more specifically on attracting attention. Metrics such as ad duration, audibility, or ad position provide information about how high-quality an ad is and thus how likely it is to generate attention. In our campaign data, we see there are strong positive correlations between the average duration of an ad and the uplift of awareness and consideration, two important steps in the user journey.

Modern demand-side platforms (DSPs) allow you to write your own bid scripts to optimize even more precisely for the relevant target values. We make use of both possibilities:

• We have designed our own attention metric that, besides pure viewability, also considers the insertion time of an ad: CPMS—the cost per thousand (mille) viewable seconds.
• We used DV360’s custom bidding capabilities to translate the CPMS into a bidding strategy. Instead of the standard optimization of cost per thousand viewable impressions, we optimize on the CPMS. The pure number of impressions may be smaller, but the time that the ads are effectively visible is longer. Initial results show we have around 30 percent lower costs with this bidding strategy than with the standard strategy.

To complement three initiatives inventory hygiene, targeting, and media buying, we recommend accompanying campaigns with digital market research, whenever possible, to validate relevant business objectives. To this end, we have designed a framework that allows for structured and early consideration of various market research tools. Thus, uplift studies on different platforms (YouTube, Facebook, etc.) are planned and integrated in line with the campaign objectives.

In addition, we continuously analyze the brand search volume on Google to gather further signals for unaided brand awareness, for example the number of people who recognize a brand or product without further reference. All these results and signals are then aggregated and condensed into an overall campaign result, which provides information about the effectiveness of the implemented measures.

We have presented various approaches based on privacy-safe data and thus managed entirely without third-party cookies. The individual approaches are individually useful and recommendable, as they all contribute to a qualitative improvement of the campaign. However, we expect the greatest impact when the initiatives are combined at the different levels and used in their fusion. So, if a campaign is targeted on high-quality inventory, targeted based on publisher data, or dynamically played out based on hard data, it will continue to achieve excellent results absent third-party cookies, especially when delivery is simultaneously optimized for attention metrics. We are convinced that the positive results will also remain visible and measurable through the accompanying implemented digital market research.