Creating a card-ranking algorithm for Hallmark

With a recommendation engine that increased the conversion rates for many of the company’s greeting card categories
Creating a card-ranking algorithm for Hallmark

The challenge

As the oldest and largest manufacturer of greeting cards, Hallmark Cards is the undisputed Goliath in the greeting cards industry. However, with thousands of daily website visitors, the company was looking for how best to leverage its data and capitalize on its growing online presence. Hallmark contacted Eraneos to help it do exactly that. The main challenge here, however, was twofold. Firstly, we needed to find out how to offer the best product to customers which are, in large part, anonymous until the very end of the purchasing cycle, and secondly, we had to figure out how to manage the daily or weekly assortment of cards in a meaningful way.

The approach

The key to cracking the first challenge of this project was to understand which card should be shown first to a potential new customer, and which would be most appealing. We solved this by modeling the uncertainty of a card’s performance. This allowed us to balance the use of known “bestselling” cards and uncertain but potentially “good” cards.

We solved the second challenge by building a “recommendation engine” that automatically ranked the cards. Here, we made sure to keep track of how many people viewed a card and how many of them ended up buying it. This allowed us to create a statistic model for each card based on its performance.

The result

As a result of the engine we built, Hallmark’s conversion rates increased for many of the categories of greeting cards and most people are now able to find their cards substantially quicker than before. Products are now also managed and automatically ranked, daily.