Testing digital concepts in a real-world setting: In-store personalization

In today’s digital interactions, businesses use data to personalize and tailor customer experiences to almost every individual. A recent study shows that 71 percent of consumers expect personalized interactions from brands and businesses, while 76 percent are unsatisfied when they see content that is irrelevant to them.

Data about customers and their touchpoints is easy to gather in a digital setting. This data is the foundation of recommendation engines, personalized marketing campaigns, and next-best-action use cases, to name a few. Transferring these capabilities into the physical world—such as brick-and-mortar stores—is challenging. You need more than just code; hardware, wires, and more are also necessary. Think about this: while you can derive the level of interest for specific products in your ecommerce store by the number of page views, how would you measure this in a physical store?

In this blog post, we will talk about the concepts and components typically found in a digital world and how they can be transferred into the physical world. Though we use personalization as our main example in this post, this approach can be used for any other capability in an ecommerce website, like churn prediction, for instance.

The component view

A typical ecommerce website has several functional areas: a searchable product catalog, a shopping basket, payment options, and product recommendations, to name a few. As an example for this blog post, we will focus on personalization and the components that are necessary to provide a personalized experience. Though there are marked differences in the implementation of ecommerce and brick-and-mortar stores, on a functional level, the components are identical. Let’s have a look at the digital world first. Afterward, we will map the components onto the physical world.

1. User identification
To deliver a personalized experience, you first need to identify your customers. In digital settings, this is achieved by allowing the users to create accounts.

2. Historical data
We need to understand which products customers are buying and interacting with. In ecommerce, this data is generally found in the purchase and interaction data. This data is crucial for any personalization use case.

3. Continuous data capturing
Though the information on actual purchases is interesting and valuable, the path that led to a purchase is just as interesting. The shopping behavior of customers is influenced by various factors: season, mood, personal preference, and more. In order for a personalization engine to capture the most current interest of shoppers, we need to continuously capture interactions with products as well as the path toward the buying decision.

4. Presentation
Lastly, we need a way to present the personalized content to a customer. In a digital setting, you can present the personalized product suggestions or personalized content in the ecommerce website.

Figure 1. Key functional components of personalization in ecommerce websites

Moving to the physical world

We have now enumerated the key functional components for online-personalization use cases and how they are generally implemented in ecommerce websites.

As we shift the focus from the digital setting to the physical world, we need to relate digital concepts to physical ones. If you compare shopping journeys, you can draw parallels between products on a physical shelf and product pages on an ecommerce website. Continuing the process of mapping approaches from the digital world into the physical world, one approach is to display the recommendations the customer sees on the ecommerce website onto a physical display in the store.

Figure 2. Smart shelf that shows personalized product recommendations to the customer standing in front of it

Following our initial categorizations of the functional components of digital personalization, we can now map these respectively onto the shelf in a physical store:

1. User identification
Usernames provide digital identification for many ecommerce websites. In the physical world, a shopping app that contains the customer’s loyalty data can be used to check in to the store by scanning a QR code at the entrance. This active opt in, combined with Bluetooth sensor data from the customer’s smartphone, will identify the user during shopping.

2. Historical data
In a digital ecommerce setting, the user account of the customer tracks purchases. In the physical world, the data necessary to provide meaningful recommendations can be sourced from loyalty programs that the customer uses.

3. Interaction tracking
As interactions in ecommerce websites are tracked digitally based on clicks on product pages, interactions in the physical world can be tracked by distance sensors as well as weight sensors on the shelves.

4. Presentation
While ecommerce websites provide a self-contained display in the digital world, we need a way to present data in the physical world. This can be achieved by installing displays for advertising or using the shopping app and notifications on the customer’s smartphone.

Figure 3. Key functional components of personalization in physical stores

Conclusion

Transferring concepts and components from the digital to the physical world can be challenging and costly. The following steps can help establish these components in brick-and-mortar stores:

1. Define your expected outcome
Identify the customer’s needs. Make yourself familiar with the customer journey and how a modern physical store can improve the customer experience and ultimately boost business. Are you planning on increasing customer engagement? Would you like to expand the average basket size of your shoppers? Setting a scope is necessary to understand what you need to do next. Think big; start small. Build and verify your assumptions with a minimum viable product (MVP), and improve from there.

2. Map building blocks from the digital world to the physical world
Think about the building blocks that you need in a digital world to achieve the expected outcome from step one. Is there a need to identify shoppers? Do you need to present something new to your customers? Map these components in a physical setting and identify what hardware and software are required to make it work in the physical world.

3. Measure, measure, and measure again
Metrics and key performance indicators (KPIs) are easy to gather in the digital world. To analyze whether a real-world solution is performing properly, it is crucial to set up metrics and KPIs. You can start by measuring the interactions with the physical solution. Try answering questions like, “How is this affecting my revenue or average basket size?” You can also place feedback buttons on the MVP to actively gather customer feedback.

Amazon Web Services (AWS), with its on-demand and pay-as-you-go nature, supports rapid innovation and development processes to quickly verify your assumptions by deploying MVPs and prototypes to the field. Go build!

Learn more about customer engagement solutions for retailers on AWS.