ClerkCore Technology

The ClerkCore uses next-generation artificial intelligence to end the assumption that 'one size fits all' for every consumer.
Good News for Consumers, e-Commerce Technology has evolved.

Beginning with Collaborative Filtering

Collaborative Filtering assumes that if Person A buys the same product as Person B, then A is more likely to buy other products which were also purchased by B.

The big drawback of Collaborative Filtering is that it is extremely demanding in terms of data, and often requires products to be sold in their hundreds before any reasonable predictions are available.

This means that Collaborative Filtering is defective in smaller shops, particularly because it can not fully recommend, and find relevant recommendations for new products, or for products that are not amongst the top 10% best sellers.

Different approaches, like Content-Based Filtering and Hybrid Recommender Systems, have been developed to fill in the gaps, but only with moderate success.


Followed by Neural Networks

Artificial Neural Network is a machine-learning approach which mimics how the neurons in our brains function, allowing computerised networks to process patterns in a similar manner and effectively learn by experience. Originally coined in the mid 1980's, Neural Networks have been revived in the late 2010's as Deep Learning.

Compared to Collaborative Filtering, Neural Networks provide significant improvements in recommendation quality and precision. Amazon has used Neural Networks as the base of their recommender systems.

The big drawback of Neural Networks (and machine-learning methods in general) is that they are vastly more complex to operate, and require an extreme amount of data to generate qualified recommendations.

This means that Neural Networks, like Collaborative Filtering, can become defective (even in larger shops) because they can not fully recommend, and find relevant recommendations for new products, or for products that are not amongst the top 10% best sellers.


The Arrival Of ClerkCore

The ClerkCore is the proprietary algorithm behind, and has been built to improve on the drawbacks which Collaborative Filtering and Neural Networks are caught by.

Starting as a research project in 2010, our Data Scientist conducted the first live experiment to compare results against both Collaborative Filtering and Neural Network installations. The results were overwhelming - 143% improvement over Hybrid Collaborative Filtering, and 57% improvement over Neural Networks when measuring in additional sales achieved!

The uniqueness of the ClerkCore technology lies in its extreme data efficiency, and this is what powers today! Our ClerkCore can yield the same quality of recommendations as Neural Networks can, but only require 10% of the amount of data.

In practice, this means that can bring recommendations of an incredibly high quality to the full range of e-Commerce users, from smaller webshops to demanding Enterprise organisations. This gives you the power to display your full product catalogue, including niche items and hot new arrivals.

ClerkCore powers the personalisation features across all of's products, from the ranking of the Search Results to the Segmentation of Customer data.

Learn More