CHALLENGE

In the course of the ongoing digital transformation of the media industry, the client wanted to further intensify its focus on exploiting operational synergies across its digital businesses leveraging the power of data. The starting point of the big data strategy was hence to exploit the economic value of the existing wealth of user and behavioral data across the digital business portfolio.

SOLUTION

In collaboration with a global consultancy, we delivered the big data project and met the following objectives:

• Strategy and roadmap to identify and exploit synergy potential
• Proof of economic value of big data initiatives through realization of 3 pilot initiatives
• Development of an appropriate business operating model for an analytics competency

RESULT

Pilot initiatives were delivered in time and budget and demonstrated that data insights can be commercialized by improving the performance of customer targeting and sales recommendation processes. The management board has approved the recommended strategy in mid 2014. The client further contracted us for the implementation of this strategy. By early 2015 the client was able to take over people, processes and technology for the newly built digital analytics organisation.

CHALLENGE

A leading Swiss health care provider intended to increase its departments’ digital awareness and the willingness to invest in a digital strategy and digital marketing in particular. The idea was hence to show that more web traffic results in more department visits (i.e., web traffic is a leading predictor of department visits) and ultimately in higher sales.

SOLUTION

After identifying and ingesting behavioral customer data from the client, we applied advanced analytics to meet the following objectives:

• Model to test for a significant relationship between web traffic and department visits
• Experiment with internet campaigns to test for a causal link between web traffic and department visits
• Model to predict department visits a specific number of days in advance

RESULT

By testing different models, we delivered the required evidence and were able to predict department visits seven days in advance, with a average prediction accuracy of over 80%. As a next step, we will start an experiment to deliver evidence for causality. Our delivered results will then be used to guide the new digital strategy.

CHALLENGE

Having over 4 million stock items in its product catalogue, the client faced the challenge of finding the ideal product-selection and -placement on their web shop. The 13 category managers were constantly fighting over the limited advertisement space in order to promote their products and reach their sales targets. Product selection and placement was done on personal gut feelings rather than on a data driven approach towards a company wide target.

SOLUTION

We iteratively developed a predictive model which generates an optimal product selection and placement for the web shop:

• Fully automated selection of products based on predicted profits (2 week forecast horizon)
• Product selection maximizes the combination of internal interests (highest margin) and customer interests (highest discounts) resulting in maximal profits
• Semi-automated product placement recommendation for the homepage, as well as the category key landing pages for the category managers

RESULT

After a successful two month pilot phase, testing 3 shop categories, the prototype went into production at the end of 2016 and became a vital means of optimizing the shop on a daily basis with all 13 categories. The solution is constantly monitored and further improved for the different needs of the various product categories, spanning from fashion to food and to multimedia.

CHALLENGE

The client provides in its online shop a full multimedia range of more than 6 million products.
Each month, more than one million customers use the online shop via the mobile app. The creation of long-standing client relationships shall be enabled by “inspiring” customers with segment specific product curation. So far, traditional agencies failed to find business relevant segments.

SOLUTION

We provided a fully data-driven approach (cluster analysis) to segment customers based on their purchase history (user centric) instead of the existing product groupings (internal view) or prejudiced personas:

• 32 business relevant customer features were identified (product categories and demographics)
• Cluster Analysis was conducted via various machine learning algorithms: K-Means Clustering delivered best results both in speed and quality
• 9 clusters were identified with ideal properties: Business relevancy, comprehensible characteristics and reasonable separability

RESULT

The final clusters showed two new segments no traditional agency was able to uncover. The client will use this clustering model henceforth for their web shop personalization. The customer cluster assignment will be updated and monitored on a weekly basis.

CHALLENGE

The digital economy and its data-driven business models are on a journey towards trust-based customer relationships. The most critical and fragile enabler for this transformation is online trust. In a world where privacy is withering away like ice in summer, understanding and building online trust are essential business capabilities.

SOLUTION

Trust is the result of a deliberate calculation process. Customers screen signals from online vendors and decide whether to take the risk to get vulnerable to the actions of another party or not. Generally, these trust cues are visible and actionable.

We combine findings from socio- and neuroeconomic research with our own experience into a framework that not only allows an understanding of how online users actually exhibit trust but rather identifying where businesses can harness and build on this trust.