With a focus on exploring innovative new opportunities that could disrupt the way AB InBev approaches online retail, it is essential for the team at ZX Ventures to understand — and to be able to anticipate — the wants and needs of its consumers.
Following our ZX.YCN event in Brussels earlier this year, and to find out more about how ZX Ventures is leveraging consumer data to deliver exceptional e-commerce experiences, we caught up with European Marketing Manager of Disruptive Growth, Julie Veryser, and Data Scientist, Jason Hall.
How are data and machine learning helping ZX Ventures to better understand customers?
Jason: From a shopper behaviour perspective, my primary focus is on using data to better understand customer needs and wants, both now and in the future.
Every individual is unique, with different wants, tastes, personalities and styles. One of our key focusses in our craft beer e-stores therefore is personalisation — personalising our sales and marketing strategies to ensure that we are delivering the most relevant and interesting content to the individual consumer.
With the growth of e-commerce there is a wealth of data. The challenge is how do you leverage the latest technological developments in big data and machine learning to really understand your individual customer?
Can you share an example of how you’ve done this?
A simple example that highlights this quite well is a feature that we ran on our French e-store, Interdrinks, where our online beer expert Christian could help you create a mixed case of beers. The concept is essentially ‘Tinder for Beer’, where customers are shown a series of products that they can either like, dislike or see more detailed information about — and the liked products are then added into a mixed case of beer.
On the surface this was quite a simple user experience, but in the background we were constantly working on optimising the customer journey by personalising the path of products that each individual customer was shown with a combination of machine learning, AB testing and probability modelling.
To generate the product paths that were shown to the individual customer we used various machine learning recommendation and classification algorithms, based on previous purchases and customer likes and dislikes. These focussed on the relationships between products and on answering various questions about the individual shopper such as:
— Is the profile of the individual customer related to the profile of another customer or a group of customers?
— Is the customer looking for a specific style of beer or wanting to explore different styles?
— Is the customer loyal to certain favourites or potentially looking to try some rarer, more exclusive beers?
Given the diverse range of craft beers that are available on Interdrinks, there were thousands of potential product paths that could be shown to a customer. As a result, we continuously ran multiple AB tests to split site traffic in order to test the conversion rates of different strategies. We also used Bayesian probability modelling to understand the impact that product positioning (i.e. the order that a given product is shown to the customer) has on product sales, as excluding this factor can show how well a specific product is performing, regardless of position.
Essentially, what it boils down to is using the latest technologies to analyse data in new and different ways, in order to constantly improve our understanding of the customer, and use this understanding to deliver the best user experience possible.
How are you leveraging data to inform decisions and strategy, and what’s the biggest challenge you face in this area?
Jason: Whether or not an organisation can fully leverage data depends on how well they build their data infrastructure.
Our e-commerce business unit at ZX Ventures is focussed on two main areas, e-retail (where we work closely with various third party online grocery partners), and our craft beer e-stores such as Beer Hawk and Interdrinks – with our learnings from one area informing the other and vice versa.
From a data perspective, probably our biggest challenge is how to deal with these multiple fragmented data sources. One of my main projects at the moment is designing a common data model to enable the organisation to view and analyse data from each of our global markets within one model.
The danger with fragmented data is that teams within businesses (especially teams in different geographic locations) end up working on similar problems but in silos. Having data sources consolidated into one data model has provided a number of advantages. Firstly, it makes it much easier for the business to share and leverage best practices; and secondly, it has allowed our data team to be structured with a global focus. By this I mean having centralised capabilities in specialised areas, such as machine learning and data engineering that can be applied globally.
In recent years there has also been a big shift in the perception of data. No longer is it solely the responsibility of the analytical functions in organisations to find insights in data. To maximise the value of data as a competitive advantage, organisations need to be able to democratise it. This means giving key business decision makers (both senior stakeholders and on the ground decision makers) access to the most relevant and up-to-date information in an easily understandable way. As a result, data becomes ingrained in all key business activities and enables decisions to become data led instead of data supported.
What do you perceive as the main differences in the way consumers shop online and offline?
Julie: While we see a lot of similarities with online and offline shopping behavior, we also see some key differences. These can be summarised in three consumer benefits that online provide: assortment availability, convenience and value transparency.
When our consumers are shopping online, they expect to find everything and anything they need, because firstly there isn’t the physical constraint of a shelf, and secondly the internet should have everything. In addition, consumers expect to be inspired online. Within beer, we not only see over indexing with our larger packs, we also see categories like non-alcoholic drinks, homebrewing, gift packs and subscriptions working very well, due to that consumer behaviour.
Consumers also expect online to deliver greater convenience, which is especially important for a product that is heavy to carry, like beer. Therefore, they are becoming more and more accustomed to fast and cheap/free delivery online.
Finally, consumers find value in the transparency that online offers. Price comparison can be done in an instant, but they also find value in the online service — whether that is recommendations via a retailer, or adding a digital shopping list to to your basket, or shopping from your past purchases.
People are increasingly shopping more seamlessly across channels — how does this impact the way you target them?
Julie: We consider e-commerce to be any purchase which has had a tech touchpoint somewhere along the journey, whether that’s shopping via one of our partners’ sites, or someone checking their phone while they’re in a supermarket.
As such, we are fully aligned with an omnishopper channel approach and look forward to working with our partners to understand how we can leverage that data within targeting. Tools like Sociomantic — which provides programmatic solutions — are great, because they allow us to understand the effect of a targeted marketing campaign to both sets of shoppers, which allows for correct understanding of ROI.
Looking ahead, what do you see as the biggest opportunities for ZX Ventures within digital retail?
It is such an exciting time to be working in this space, especially at a European level because each market is so distinctive in its e-commerce approach.
The opportunities I see, besides the omnishopper, are threefold, depending on the perspective. From a supplier perspective, it would be the rise of tech cross-platform solutions like Criteo, which allow us to further our digital tracking of our media investments and allow us to activate in a scalable way. The digital world should be a fast-paced environment with real time optimisation, such as with Facebook and Google so the opportunity is for this to be within retail. From a consumer perspective, the opportunity lies in an even stronger acceleration of convenience, whether that means faster delivery across the board or even delivery in home when the consumer might not be present. Finally, from a retailer perspective, the opportunity lies in closer collaboration and transparency with suppliers — once we are able to share our learnings seamlessly back and forth, there’s no stopping digital retail.