Intro
Our story with Hebe begins back in 2012 when it was just a small drugstore chain with only several stores. A decade later it is one of the biggest drugstore brands in Poland, with 291 stores (and a plan to open 30 more in 2022) and thriving e-commerce. We have the pleasure to accompany Jeronimo Martins on their way from a fledgling business to one of the most important players in the health & beauty drugstore market. All those years later, we’re still by their side to help them grow further by perfecting marketing activities, developing data warehouse, and delving into data to understand its customers’ behavior even more.
Loyalty program
- 5 million loyalty cards issued in the last 10 years
- 2.9 million active loyalty cards
- My Hebe in 2021: +13%
Hebe reached out to us for the first time in 2012, not even a year into their debut, to launch a loyalty program as a way to celebrate its one year on the drugstore market. In just a couple of weeks, we successfully planned and implemented the loyalty program: we designed the paper (yes, only paper at that time) loyalty card, planned the process of obtaining the card along with the form to gather necessary data about loyal customers and reached out to Hebe’s POS supplier to enable us to track transactions of the loyalty program members. This allowed us to obtain a lot of data to analyze and play with to gain valuable insight into customer purchasing patterns. We were able to match them with the demographics of the customers to draw conclusions and implement the right marketing strategy to help them grow and get more members. Throughout all these years, Hebe has, so far, issued more than 5 million loyalty cards, with almost 3 million loyalty cards active by the end of 2021. In 2021 alone, My Hebe increased by 13%! According to Loyalty Insight Report 2022 examining loyalty programs in Poland, Hebe ranked second as the drugstore loyalty program with the highest engagement and satisfaction in 2022.
Diamonds - loyalty program rewards
- rewards program based on gamification and complex calculations – increasing customer engagement
A couple of years ago, Hebe wanted to expand their loyalty program and offer rewards to increase frequency and loyalty, and we were all in right away. The difficulty was to find the right amount that should be spent by a customer in order to collect the reward and that, also, wasn’t too hard to reach for an average loyalty program member.
What we did was prepare complex calculations to find an optimal spending threshold (based on quartile spend and probability) that would both encourage customers to increase their basket size, make the program fair for all members and ensure that the reward program is profitable for Hebe in the end. In effect, customers below the spending threshold have an incentive to shop more to get the rewards and those who already spent more than the threshold are rewarded either way. And that is how My Hebe VIP was born - you get a diamond each time you spend at least 25 PLN and once you collect 10 diamonds you automatically become My Hebe VIP for the next 90 days. My Hebe VIP status gives you extra vouchers, free delivery or early access to certain promotions. But it doesn’t stop there, in those 90 days, you need to collect another 10 diamonds to keep your status active for the next 3 months, and so on. Once you get access to special offers, you, hopefully, want to continue having your VIP status if the rewards are worth it. The program has been active since its inception and has introduced fun gamification in our loyalty program that increases customer engagement and improves sales.
Data warehouse and reporting
- data warehouse migration to AWS Redshift - increase in efficiency (100x)!
- processing coupons in real-time
Creating a loyalty program in the retail space always starts with data scattered in different systems. From the very beginning, we were obsessed with bringing them all together in one place. Over the years, we’ve been gathering customer data, campaign interactions, sales data, and price change history. Having all that knowledge at our fingertips gives us a complex and very specific overview of customers and their purchasing patterns. However, as Hebe was expanding, its constant data growth decreased the efficiency of the warehouse we built (at that time we were using IBM DB2), so we came to the conclusion that moving to cloud is the best thing we can do. Our migration to AWS Redshift was crucial for us and for Hebe to store more data and to increase the efficiency of the queries, but also protect us from technical debt. The migration was an enormous undertaking and it took us two years, countless hours of work and was a spectacular success in the end. Hundreds of tables and automatic processes were migrated, but most importantly, it was smooth and Hebe and their data did not suffer at any point. Before the migration, reloading the data warehouse used to take around 4 hours, now it’s only 8 minutes! Thanks to our data warehouse, we are able to process coupons in real-time allowing us to send a coupon to customers within just minutes of them signing up for an account in My Hebe. Through our warehouse, we are also responsible for creating promotions in cash registers (individual promotions). That’s why centering all our activities for Hebe around a reliable data warehouse was a must and continues to be a focal point for all of our activities.
Marketing automation
- omnichannel marketing automation
- over 40 triggered campaigns
- automated emails, text messages and push notifications
At the beginning of our cooperation with Hebe, we decided to implement marketing automation to send automated emails and text messages with latest offers to members of the loyalty program. Even though these two activities might seem easy to do, as Hebe’s customer base was quickly growing, it was complex enough for us to introduce IBM Campaign (earlier called IBM Unica) that helped us automate it fully. In 2016, however, due to lack of updates and further development of IBM Campaign, we decided to change the marketing automation platform to Selligent. As Hebe was expanding every year, there were more channels for us to take advantage of and this move to use Selligent gave us countless possibilities in terms of the scope of services we could offer. We could continue to help them grow and boost customer engagement as well as improve frequency and basket size. Apart from text messages and emails, we could now also automate push notifications, and later, when Hebe became fully omnichannel by entering e-commerce market in 2019, create really precise triggered campaigns (now over 40 different scenarios that help us create tailor-made communication in different channels) and introduce our marketing automation in their mobile app. By the end of 2021, their online channel reached 13% of sales after just 2 years of operating, doubling the sales from 2020. Hebe is aiming to keep its online sales growing and we’re ready to keep supporting them along the way.
Control group
- periodic control group to show effectiveness of marketing communication
To demonstrate the effectiveness of the communication with loyalty program members, we decided to, a few years ago as an experiment, create a control group. It was crucial to select a great number of people at random in order to detect statistically significant differences. After thorough analysis, we decided to select 7,000 customers who are loyalty program members and for the next 2 months, they didn’t receive any active marketing communication. We then compared the results with those who received it during that time and we were able to truly see the difference our marketing communication makes. The experiment clearly showed how much extra turnover our marketing communication with loyalty program members generated during that time. It allowed us to not only monitor our effects, but also showcase how valuable our communication with customers is in terms of sales. Something that started only as an experiment has stayed with us for good. Now, every 2 months, we create a new control group, monitor our triumphs and, happily, report them to Hebe.
Recommendation model
- proprietary algorithm responsible for recommendations, present and future ones
- fully controllable, explainable and consistent
To increase customer engagement, frequency and basket size, we needed a scoring model for repetitive purchase settings that would encourage customers to buy their favorite products at just the right time. Instead of using black box where there’s no information surrounding its internal workings, we made a decision to create our own recommendation model.
Enter Lochness model, which we fully developed from scratch for continuous, non-contractual relationship with customers based on Probability Models for Customer-Base Analysis by Peter S.Fader and Bruce G.S.Hardie. We called it Lochness not because it’s scary (far from it), but simply because of its shape. It’s based on a set of variables such as purchase history or time since last purchase, among others, and generates recommendations for every customer in every single promotion. Lochness could also generate future recommendations for customers who could be informed about upcoming promotions at POS in a personalized way. It’s consistent, available every single day and truly explainable – we control the parameters and can not only test numerous approaches, but also perfect them.
