Online shops pay a lot for “paid traffic” through various advertising and marketing channels. Nevertheless, conversion rates often remain very low. This is not only expensive but also neither efficient nor effective.

However, conversion rates increase with an optimized user experience (UX), meaning good personalization. Previously, cookies were used for this purpose to recognize a user and then personalize the offer for this user based on the past behavior during the last visit(s). But due to increasing rejection and deletion of cookies, often 60-80% of the traffic remains “unknown” and can no longer be personalized with the previous methods.

Therefore, new, innovative solutions are needed that respect the privacy of users while still delivering individually relevant content. Enabling this and thus being able to sustainably exist in the market is a major challenge in e-commerce.

The Challenges at a Glance

1. Personalization without consent and without cookies

How can online shops deliver relevant content without cookies, and thus without any knowledge of the user’s last visit to the shop? In most online shops, up to 60-80% of the total traffic consists of unknown users. Even a classic Customer Data Platform (CDP), which collects 1st party data on user behavior, is very likely to store personal data according to GDPR and ePrivacy definitions. However, capturing these data requires the consent of the user. Therefore, the problem remains that a large part of the traffic consists of unknown users – and this is increasing.

2. Real-time data activation – learning with every click

Without cookies, every recurring visit of a user is always recorded as a new visit by a new user – the user simply remains unrecognized. If the customer journey of a user is no longer recorded, only the current session of the respective user remains for personalization. Therefore, real-time data activation must occur in the ongoing session, even without knowledge of the previous customer journey.

Real-time data activation is based only on the given, often only anonymous data points of a current session and the user interactions made in that session. Thus, with each click, a real-time analysis with immediate decision “What should be displayed?” is conducted, and this decision is immediately played out. Real-time data activation without pre-made segments or prepared “personalization recipes” is a major technical challenge because it must happen in a few milliseconds, albeit a solvable one.

3. Targeted merchandizing – Analyzing individual relevance and aligning with business goals

Only with the above-described data activation can the online shop always deliver the appropriate personalization for each user in their current situation. However, this user-specific relevance should also be in line with the business and campaign goals of the online shop. Targeted merchandising must find the balance between hyper-personalization and pursuing the own, shop- or campaign-specific business goals.

4. Maximum scalability without rule sets and pre-defined segments

Due to the many different users, their characteristics and intentions, the various marketing channels, the product variety, and shop categories in combination with the desired campaign, shop, and business goals, a fully automated match of all “these ingredients” is required to personalize efficiently. Personalization should not be done for the sake of personalizing, but should be actioned with a clear business goal. Mapping these high-dimensional feature combinations with rule sets for pre-made segments does not scale. Such a manual or semi-automatic approach can rather lead to an enormous and rapidly growing effort for setup and maintenance. Instead, a scalable, data-based determination of individual relevancies for all conceivable situations and constellations is needed.

A solution approach that makes both sides happy, the customer and the online shop: Customer Engagement Platform

The Customer Engagement Platform (CEP) ODOSCOPE offers a comprehensive solution for these challenges. With its innovative approach to personalization in harmony with AI-driven merchandisising, it provides sustainable solutions for using the existing traffic in the shop and for long-term customer loyalty. Specific KPIs such as conversion rates, sales, margin, etc., develop sustainably positively; shop management is simpler, faster, and much more transparent.

How does ODOSCOPE’s CEP meet the challenges mentioned above?

1. Personalization without consent and without cookies. Look-alike-audiences.

The Customer Engagement Platform enables personalization without cookies and without user recognition. The current user’s anonymous session data points are analyzed at any click, and a dynamic look-alike-audience is determined from the overall data history. This look-alike audience, i.e., a group of users who are as similar as possible to the current user, must be large enough to show significant behavior and small enough to be relevant for the current user.

ODOSCOPE itself does not collect data, nor does it require a pixel integration. It does not need the user’s consent to execute personalization based on already collected data and the current session data. The Customer Engagement Platform is based entirely on the legally captured web analytics data of the respective shop. The real-time analyses of the look-alike audience are based on anonymous session data such as “iPhone from the Cologne metropolitan region on a Monday afternoon via campaign link xy” plus the user behavior from the current session “Category: Winter jackets, Filter: Size 40, Color Black, Price up to 200 €“.

If a user consents to cookie setting or is even logged in, then the further (personal) data can also be used for personalization.

2. Dynamic real-time data activation – learning with every click

Since the majority of users in e-commerce are no longer recognized as returning visitors, only the currently ongoing session with its data points and user interactions remains for optimizing the user experience. ODOSCOPE analyzes an individual look-alike-audience in a few milliseconds, sharpens it with each click of the user, constantly adjusts it, and thus enables the real-time delivery of the most individually relevant products. Rules passé – dynamics olé!

The dynamic real-time data activation thus adapts to the current user behavior, regardless of whether filters and categories are changed and thus the interpretation of the intention can change.

In contrast, predefined rules and pre-made segments can only address the behavior of a user from the last visit. However, a new visit by the same user can have a different intention or pursue a different goal. How often are users shown the last viewed products again or recommendations made only based on a past purchase, although the current visit has a new, completely different intention than the last visit of the user?

3. Balancing relevance and AI-driven merchandising

Through the interplay of real-time data activation with AI-driven merchandising, any user behavior can be dynamically and appropriately addressed. It requires no rule sets, no predefined segments, and no adjustments per category. The AI considers “weightings” of certain characteristics or data points (e.g., “play out a bit more sales products to users coming via Google Shopping vs. show more new products to users coming via newsletter“) and acts fully automatically.

Thus, the dynamic real-time personalization always aligns with the individual relevance for the single user and along the desired campaign, shop, and business goals. No matter through which channels a user comes or in which (changing) categories a user is browsing. Instead of often very complex rule sets, only the “hint” on a few steering factors is necessary to let the AI automatically and always data-based handle this highly complex task.

4. Scalable personalization

Dynamic real-time data activation can immediately and precisely respond to changing, new, or other behaviors and interactions of a user and adapt the user experience in real-time to the shown user behavior. Instead of acting with rule sets for all user segments and their eventualities individually, which may also need to be set up and maintained differently for each product category, the AI can dynamically analyze all variants and influencing factors. The AI can play out an individual decision with the highest possible relevance for the single user.

The fully automatic control and dynamic deliveries occur not in a black box or without control or close monitoring. To steer the system and continuously optimize it, results and developments are shown in a comprehensive dashboard and discussed regularly within a weekly jour fixe, new strategies are developed, set up, monitored, etc.

Personalization across all touchpoints of customer journey: From search result lists, to product lists, to on- and offsite recommendations, personalized display banners, …

Personalization must be thought of holistically. At all possible touchpoints where the user interacts with the online shop, the content must be orchestrated, coordinated and individualized. Only then users can experience a truly individual UX. This requires a technology that uses and combines the data of the individual touchpoints and adjusts the product displays accordingly. This coordinated offer, tailored to the user’s interests in their respective shopping situation, is increasingly referred to as hyper-personalization. Below you find touchpoints across the different phases of customer’s journey listed and described, how ODOSCOPE ensures this hyper-personalization:

Awareness phase

Personalized Display Banners: Adaptive Creative Optimization (ACO)

The Customer Engagement Platform personalizes display banners according to individual relevance. For this, the ad-server must communicate with ODOSCOPE to identify the most individually relevant banner from a number of banner-variants in the few milliseconds between winning a bidding and displaying the banner.

Previous ACO projects showed a 300% better click-through rate by playing out the most individually relevant banner: Instead of a randomized banner display or even just a single banner variant for all users, it is individually decided for each user which banner variant will be displayed – in accordance with the individual look-alike-audience of the user to whom the banner is shown. The look-alike-audience is analyzed from the data of the shop to which the banner links, ensuring the user’s personalization in the context of the advertiser and its product range. This has demonstrably led to a significant uplift in banner-generated sales and purchases.

Discovery & Search phase

2. Sorting of Search Results

Internal search often replaces navigation: users enter search terms for products and receive links to the corresponding product pages as search results. The products that are most relevant to the individual user should be at the top. This delivers an individually sorted search results page that is most likely to result in a click on the search results, sending the user directly to the product pages most relevant to them. ODOSCOPE ranks these search results in real-time according to individual relevance, thus optimizing UX in this area as well.

3. Relevance Ranking in Product Lists

On category or product overview pages, users are getting inspired or search specifically for products relevant to them. Here, it is important to list prominently the products that are relevant to the current user. Only with product lists which are sorted by individual user relevance a truly personalized UX is achieved that responds to the current user behavior and intention. With each click, the list is recalculated, if necessary, resorted, and the product offering is adjusted to the interests of the user, just like a sales consultant would do serve a client in a physical store. For the online shop, this requires real-time analysis with real-time data activation that immediately creates a personalized view with each additional click.

Acquisition phase

4. On- and Offsite Product Recommendations

Classic product recommendations are based on the statement “customers who bought x also bought y.” However, this is not very specific, as according to this rule, everyone is offered another bestseller after they have already bought one. The Customer Engagement Platform relies more on a user-centered recommendation approach. ODOSCOPE’s recommendations are based on the analysis “customers like you who bought x also bought y.” The recommendations are thus more targeted, individual, and relevant than a generic conventional recommendation that displays unspecific recommendations like “customers who bought x also bought y.”

Summary

Personalization can be thought of and implemented in a variety of ways. For some, recommendations are sufficient, while for others, it should be more comprehensive and holistic. The Customer Engagement Platform ODOSCOPE considers personalization in a holistic and long-term manner. Data is at the center of all activities. Data means knowledge that can be leveraged, utilized, and ultimately monetized. This goal is pursued with the approach of this CEP. While the long-standing experience of an e-commerce manager usually forms very good guidelines, gut decisions can no longer be the basis of business decisions due to the diversity of users, the large product range, and the high dynamics in e-commerce.

The Customer Engagement Platform ODOSCOPE provides both the online shop operator and the operational e-commerce team with a platform that acts data-centric and quickly, employs highly innovative technology, and offers full transparency and control. The efficiency of this approach speaks for itself: 100% of the traffic is leveraged and the operational staff experiences an unprecedented speed and ease in performing their daily tasks.

With foresight in planning and setting up the underlying system architecture, the platform is 100% GDPR compliant and personalizes also without cookies and user consent. This makes the platform secure, user-friendly, and future-proof.