— Is an online bookstore with an assortment of over 50 thousand books. Book24 officially represents many Russian publishers, for example, EKSMO, MIF, AST, the Russian Textbook Corporation, and their imprints: Like Book, Fanzon, GrandMaster, Hourglass, Corpus, etc.
Towards the end of 2016, the Exmo-AST publishing group merged four online stores into one. Which is how Book24 came to be. Store managers placed a task before Book24 Marketers. And that task was to ensure that readers don’t receive information regarding all books, but rather just those that they are interested in.
Book24’s main communication channel with customers is via email newsletters. Therefore, we decided to work on improving personalization. Book24 contacted Mindbox to help implement plans.
The result of our joint work together was 3 million rubles of additional monthly revenue from Book24’s email channel in just half a year.
The result was measured using the global control group method.
In order to increase the personalization level of newsletters, it was necessary to:
- Transfer data about all clients from the old CRM systems to a single Mindbox CRM system.
- Divide buyers into segments depending on their preferences and behavior in order to send content that is appealing to each segment.
- Launch new manual and triggered mailings, while improving existing mailings using Mindbox tools.
- Implement automated product recommendations.
- Use an algorithm that controls the frequency at which letters are sent to shoppers.
Book 24 Opinion
We started our work with Mindbox at book24.ru a little over a year ago. Over the last year, we managed to not only retain old subscribers but attracted a whole bunch of new ones as well. We also increased profits from newsletters.
Like many other Russian stores, Book24 moved away from the foreign Mailchimp service due to their increasing prices and difficulties in providing adequate technical support.
We were looking for a contractor with good functionality that could be customized to fit other specifics of our business, and one with expertise in email marketing and broad opportunities for database segmentation and the automation of content substitutions in letters.
What we had at the entrance: four obsolete databases of different online stores from one publishing group, which needed to be combined and introduced to the holdings new online store, which retained all the advantages of the old one, but under a more modern and convenient hood.
In addition to creating a mailing system, we were faced with the task of tripling Book24’s subscriber base in a year and to not lose anyone when transferring data from our CMS to 1C and Bitrix. No problems arose regarding this, everything was like clockwork thanks to the integration.
Over the course of the year, more than 25 new triggers were launched, including AB tests: going from most popular, for example, phased reactivation, to specific ones, such as discounts for the purchase of office supplies together with textbooks, or recommendations for more books to buy from a series. Thanks to the launch of these triggers, we managed to attract 2.5 thousand orders throughout the year, and we plan on increasing this figure.
Having completed the integration, we were now able to take full advantage of segmentations. For example, we stopped sending letters to those who just recently made a purchase Because there were times when a client would receive a letter with a description of a new promotion and would cancel an order that has already been placed but not yet delivered.
As a first step, we combined four databases of the old online stores of the publisher into one database with a single profile for each Book24 customer. Some consolidation was needed so that Book24 could use customer data for segmentations.
In manual newsletters, segmentations are used to share insights about promotions or new products to a narrow segment of customers. For example, buyers who love fiction books receive information on promotions related specifically to the fiction genre.
The conversion rate to orders for this newsletter is 0.236%.
Coming up is another example. When new products arrive at the store, Book24 marketers manually assemble the appropriate segment via the Mindbox platform and distribute the newsletter. Thus, detective story lovers find out about new detective novels just in time.
The conversion rate of this newsletter to orders is 0.777%.
Segmented triggered mailings
In triggered mailings, segmentation is used for automated email campaigns. For example, customers who make an order two weeks after a purchase, receive a letter asking them to leave a review in exchange for points.
Such letters, in addition to a primary purpose, have something else going for them. And that is to lead customers over to the site and convert them into an order. From December 2017 to July 2018, 0.402% of all letter recipients were converted to orders.
The conversion of this newsletter to orders is 0.427%.
As a result of triggered mailing personalizations, Book24 established clear communications with different segments of consumers.
Since we began personalizing mailings, the number of unsubscribe requests from all Book24 letters has reduced, compared to the control group. In January 2018, there were 0.28% unsubscribe requests, and in June, just 0.08%.
Book24 marketers use product recommendations in both manual and triggered mailings. All types of recommendations are generated automatically using Mindbox algorithms.
In manual mailings across the entire Book24 base, two types of recommendations are used: personalized recommendations for each customer segment and recommendations for the most popular books.
For newsletters with book recommendations by segment, Book24 uses preference-based separations. These are four segments: non-fiction, fiction, business literature, and children’s literature. Buyers of each segment receive letters informing them of a potential book to choose from a niche that interests them. For example, business literature fans will receive a recommendation letter for new business publications.
To send a newsletter with personalized recommendations for the entire database, marketers needed to make only one letter. A block of recommendations is automatically inserted into each letter in accordance with the segment. Three to four letters are sent out per week. Currently, 1.5 hours are saved on preparing one letter, which is 4.5-6 hours per week.
The conversion of this newsletter to orders is 0.087%.
We tested which product recommendations work best
AB tests were conducted by Book24 to figure out which recommendations work best.
Test: determine which recommendation algorithm generates the most revenue – recommendations of popular new books or personalized recommendations based on behavior.
The test showed that the average check is higher for buyers who received letters with recommendations based on popular new additions, and not based on their personal behavior.
Test: determine which recommendation algorithm generates the most revenue – recommendations of all-time bestsellers or personalized recommendations based on behavior.
According to test results, it turned out that there is no statistically significant difference in these mailings.
So, we found out that product recommendations based on behavior work just as well or a bit worse in comparison to popularity based recommendations.
We plan to conduct more tests to find a solution with an even greater conversion to order rate.
Automated product recommendations in triggered mailings
For triggered mailings, Book24 uses a series of product recommendations. If the buyer orders a book from a certain series, then five days later, the system will recommend him another book from the same series.
The conversion to order rate for this newsletter over our entire collaboration comes out to 1.074%.
«In May 2018, we decided to master the automation of product recommendations. We wanted to reduce the manager’s labor costs and increase newsletter conversions. Now our new products, bestsellers, and discounted products take into account user interests. Now we’re thinking about transferring our commodity and promotional feeds to Mindbox, so that our banners and promotions are of maximum interest to users».
Ekaterina Mosyaga, Head of Direct Marketing
Personalised the frequency with which subscribers receive letters
If subscribers receive emails too often, their interest tends to drop. If done too rarely, brand interest will drop. In order to find the optimal frequency of newsletters, we use the newsletter sending frequency control.
This is an algorithm that monitors customer behavior in the database, analyses their actions and calculates the best interval under which to send out newsletters, so as to not bother people.
If a buyer stops opening mailings, then the frequency decreases. But if he actively reads them and clicks, it increases.
Results provided by the algorithm for Book24:
- A decrease in unsubscribers for manual mailings from 5.62 to 3.79%.
- Open rate increase from 6.73 to 11.64%.
- The number of click-throughs increased from 0.95 to 1.78%.
It’s worth noting that at the same time, the number of letters decreased by half, yet the audience coverage did not change
«In December 2017, Mindbox suggested that we try a new tool that would allow us to control the frequency at which emails would be sent. Initially, a decrease in the frequency per single user within the bounds of the test group led to a decrease in revenue. But we carried on with testing in order to find the optimal number of letters to send per user without taking on a loss in revenue. At the same time, I was very pleased with indicators such as the Open and click rates during our test phase».
Ekaterina Mosyaga, Head of Direct Marketing
Personalisation with benefits for the client
Book24 wanted to make the process of choosing books more convenient for customers. Therefore, its marketers increased the level of personalization in their emails via segmenting, sending recommendations, and controlling the frequency at which emails are sent.
As of July 2018, personalized mailings bring the company, an additional 3 million rubles of monthly revenue, on average.
The Mindbox Team
You will be interested in the following materials
- VamSvet: how to achieve an ROI of 1135%
- How United Colors of Benetton launched an omnichannel loyalty program based on machine learning and cross-channel communications with product recommendations
- Sending newsletters with AMP content: acquiring permission from Gmail and Mail.ru
- Book24.ru: story of an email channel share doubling in half a year