Why is "no cookies" not the same as "cookieless"?

Matthias Bettag
Sr. Data Strategist
May 4, 2024
5 min read
Share this post

The move towards a "cookieless" future does not signal the end of all cookies but rather a significant shift in how cookies are used and the types of data they collect.

What led to the term “cookieless” is anything but the end of cookies. The post "Cookies are delicious and I just can’t get enough" of Julien Coquet enjoyably explains the different cookie types and  processing over time. Indeed, the term “cookieless” is misleading – a rogue who thinks it’s just a marketing rewording to overcome the bad image of cookies..

There is a significant difference between the terms “no cookies” and “cookieless”

"No cookies" means that no cookies are needed to operate (which applies for the Customer Engagement Platform ODOSCOPE. “Cookieless”, instead, is a common buzzword in the mar-tech industry for avoiding third-party cookies, but first party cookies are still needed and involved.

In fact, almost every mar-tech tool relies on cookies in order to recognize a user, track their behavior and their customer journey, provide next actions in order to display individually tailored offers, wordings, views, recommendations, vouchers and so on – but not now, only once the user visits again.

On the other hand, the amount of users who either reject or delete cookies is growing. The rate of “unknown” users in ecommerce is often up to 60% and more. More than half of the incoming traffic cannot be operated by any cookie-based tool:

  • Almost all mar-tech tools for optimizing conversion rates, improving user experiences and personalization require cookies.
  • More than 60% of the incoming traffic cannot be handled by any cookie-based technology.

Is there an option to optimize for the entire traffic?

There is good news: Yes, there's.

It is much less complex to integrate as you may think. And you already have all the data which is necessary. ODOSCOPE takes a modern approach which overcomes the old-fashioned way of more-or-less static optimization process by segmentation and with predefined rules. See the illustration to better understand the different approaches:

Graphic showing the risk of cookie-based personalization.
Figure 1: Cookie-based personalization

As you can see in figure 1, any cookie-based personalization is vulnerable to the user acceptance of cookies. Many personalization engines do not act user-individually but by behavioral segments. Also, what is displayed by a personalization engine (which can only happen up from a recognized 2nd visit) is often rule-based. The more predefined segments and the more use-cases are existing, the more rules have to be defined and maintained. This can become a severe bottleneck for scaling any personalization.

This approach can be seen as “faster horses”, to re-use Henry Ford’s famous words, but it is still a rather static methodology which is focusing on viewed products or single KPIs, but not on the user theirselves: It is not identifying user individual patterns and likelihoods independent from last visit’s behavior, and it cannot display user-individual preferences dynamically and based on the actual behavior within the current session.

The next illustration shows ODOSCOPE’s automated, dynamic approach which works with in-session data (and optionally more data sources) and in real-time. It is not relying on cookies, but of course it can use information from cookies and logins when existing and available: 

Overwiev of dynamic user-individual personalization in real-time. ODOSCOPE.com
Figure 2: Dynamic user-individual personalization in real-time by ODOSCOPE

Three differences between the two approaches

Data lakehouse

The first important difference between the two approaches is the data lakehouse: Instead of analyzing data in each silo (advertising tools analyzes user behavior from users who clicked on an ad, recommendation engines analyze user behavior from users who bought an item, etc.), it analyses all different data points from all different data sources altogether. This identifies all significant relevancies across all data sources. For example, users with an iPhone from a metropolitan area on a weekday afternoon may buy significantly different products than users who are visiting on a weekend on a desktop computer from a rural area (every shop has different significant “data audiences” which allow to be acted on).

AI-modules' real-time capability

The second difference is the AI-modules' real-time capability: The highly complex correlation analyses must happen automatically and very, very fast and for every user click individually, it takes less than 20ms. The result is the dynamic identification of the current user’s look-alike-audience and the audience’s significant (product) preferences. This is a purely data-driven and holistic approach which allows real-time data activation based on the actual behavior and in combination with any other data source, e.g. CRM, ERP, customer journey, and even weather data or other external factors. The AI-modules are optimizing product list sortings, recommendations, search result sorting, individual newsletter and even display advertising.

Flexible and dynamic “understanding”of what happens just now

The third difference is the flexible and dynamic “understanding” of what happens just now: The real-time analyses provide the most likely most relevant products for every single user in the current moment and based on the current user’s data points and behavior. ODOSCOPE’s API (also by plugin) communicates with the shop system about which products or elements have to be displayed for this very user.

The data activation does not necessarily require any consent: Even a user who completely denies consent, including the most simple web-tracking, would still come at a certain time of a certain day and the user agent contains the device type (otherwise no mobile view could be displayed for anonymous users) and other data points. On top of that the user clicks on categories, products, filters, etc. ODOSCOPE is completely stateless, it analyzes the existing in-session data versus the data history to identify and display what is individually relevant. Yet, nothing is stored and no additional user information is recorded.    


Approaches like the one of ODOSCOPE are a new methodology for personalization and conversion rate optimization. They are not a “faster horses” approach. ODOSCOPE combines every given data source, no matter if anonymous or user-related (when consent is granted or after login) and including external data sources like weather, traffic, events, etc. And it includes an assortment control , i.e. merchandising to align the business goals with the user preferences.

The real power comes by it’s speed: Other than common customer data platforms ODOSCOPE can analyze very large data sources in real time (<20ms) and evenly enables real-time data activation towards the user while the landing page is rendering and with every further click. This all is entirely based on existing data. No new pixel has to be implemented, no additional consent is required (as the underlying data has been tracked with granted consent), and it works also for new users, users who deleted cookies and users who do not even grant consent. If we would have been invented the word “cookieless” we would have meant it as it sounds. However, ODOSCOPE can work in parallel to cookie-based solutions with focus on unknown users, as well as a full blown customer data platform - but with dynamic and flexible real-time data activation capabilities.

Share this post