Original version published on Marketingfacts.nl (Dutch)
January 5, 2015 - Many studies prove that personalizing your web shop and your e-mail campaigns where customers receive relevant content are converting higher than when using generic content. But personalization is easier said than done. Especially when you only offer one type of product. Wijnvoordeel.nl, the largest online wine retailer in The Netherlands, was faced with this challenge. How do you offer wine that is a perfect fit with the individual customer?
It seems logic to offer only red wines to someone that has a preference for red wine, and offer white wines to people that prefer white wine. But the world of wines has many other characteristics that make this method of recommendation less effective. For example a wine from Chili has a different character than a wine from France from the same grape. There are also differences between quality, taste profile, and price as well as grapes and years.
Jan den Bakker, e-commerce manager at Wijnvoordeel picked up this challenge and developed a strategy to increase the conversion by assisting the visitor choosing the perfect wine. He chose to use Pleisty’s recommendation engine to automate this task. Pleisty’s personalization tool tracks all actions from the visitor (clicks, views, purchases, etc.) and stores these in a personal profile. The tool then compares this data with all actions from other visitors to determine in which items the visitor is interested at that moment, using automatic algorithms.
Personalizing e-mails and web shop using Big Data
For advanced personalization you need data: lots of data. Algorithms calculate in real-time which items are of interest to the visitor. For example, in a wine shop you will notice more demand for a fresh rosé on warm summer days, while rich red wines are more popular during the colder months. A personal recommendation that is calculated in summer might not be applicable in winter. The Pleisty personalization tool that Wijnvoordeel is using calculates continuously in real-time what the proper recommendations are for the individual visitor.
These real-time recommendations are then shown on web site and in e-mail marketing campaigns. The more the visitor visits the web site, or responds to e-mail campaigns, the better Wijnvoordeel can recommend relevant items for the customer.
Difference in personalization on the web site. The left image shows the recommendation for someone interested in Italian wines. On the right are the recommendations for someone that is interested in Rich and Tasteful Spanish wines below €10.
No recommendation in e-mail when product was purchased
How perfect a recommendation seems for a web site visitor, sometimes you should avoid that the visitor sees a certain recommendation to prevent buyer’s remorse. You can imagine that a customer doesn’t like to see the wine he just purchased, now offered with a 30% discount. Wijnvoordeel has defined a few merchandising rules that are built in the automated recommendation algorithms.
These merchandising rules can define which items the tool should exclude, as well as which items the system should show. For example in automated abandoned cart campaigns and cross-sell campaigns. When someone has purchased an item in the past 7 days, Pleisty will show the relevant cross-sell items but the purchased items are not shown. Has the user added products to the cart without purchasing, then these items are shown on web site and in e-mail.
Adding personalization widgets to web site and e-mail marketing templates
Space is reserved on the web site that is filled automatically and dynamically with personalized recommendations via a web service. When the widgets are inserted on the web site and the algorithms are defined, the system is self-learning and rarely requires manual intervention to show relevant product recommendations.
A similar process applies to e-mail marketing. Pleisty pushes the proper recommendations to the e-mail profiles using the API of the e-mail marketing software. The e-mail marketeer creates dynamic fields in the e-mail templates that are linked to these database fields, and the e-mail marketing software is generating and sending the e-mail messages. This method ensures that e-mail marketeers will keep full freedom for design and content of the e-mail, and can - for example - combine static content such as promotional banners with dynamic content such as personalized product recommendations.
On the left the fields in the database of the e-mail marketing software (Copernica) that Wijnvoordeel is using. On the right a description how images, descriptions and hyperlinks are inserted in the e-mail.
Immediately after the initial implementation Pleisty will contribute significantly to an increase in profitability. The number of orders per client is 58% higher and the conversion is 17% higher for users that have clicked on personalized recommendations, compared to users that did not. The Average Order Value (AOV) for orders that contain recommended items is 32% higher compared to orders without recommended items.
These numbers confirm to Wijnvoordeel that their personalization strategy is effective and they will continue on this path by adding personalized trigger-based campaigns and experimenting with other personalized techniques. One of these experiments is the personalized sending time. What is the best time to send an e-mail to an individual subscriber? Wijnvoordeel will also test the effect of personalized subject lines (“Dear Andrew, these Spanish red wines are selected for you”).
Currently Wijnvoordeel sends 3 newsletters per week, of which only one is personalized, and two are special promotions. Activity-based campaigns will be added soon, and will include an abandoned cart campaign, recently viewed campaign for visitors, and cross-sell and repeat purchase campaigns for customers that purchased.