View all articles

How we deliver next-gen retail experiences

By Stefan Gerber

In today's highly competitive retail landscape, staying ahead of the curve is crucial for success. With more customers using online delivery services for their groceries, it's essential to leverage the right technologies and strategies to generate value and drive growth.

In this post, we'll explore how a graph data management and intelligence-based approach can bring immense value to retailers seeking to provide a next-generation experience for customers, reduce short-lived stock waste, and introduce additional streams for revenue generation to their platform.

Let's begin by examining a simplified retail data landscape, which includes suppliers, stores, products, customers, and orders. The product lineage flows from a supplier to a store to a customer and is recorded as a completed sale by creating a new order and assigning it a unique ID.

A simplified entity relationship diagram (erd) depicting different business units within a simplified retailers supply chain.

Each of the above tables consists of records for each individual entity or unit (much like rows in a spreadsheet) and is described by properties (columns in a spreadsheet).

These different entity tables are connected by dependencies, as illustrated by the coloured lines connecting them. A dependency comes into play when, for example, we need to link an order back to a store. In that case, we would first follow the green dependency to look up the product associated with the order and then follow the purple dependency to look up the store associated with that product.

Just to reiterate, this is a very simplified example. Large retail organisations have to deal with thousands of relationships and tables containing billions of records. This methods of joining tables to retrieve records is perfectly fine and has been a tried and trusted method of record-keeping for more than four decades. But…

We live in an age of innovation. Customers expect personalised experiences #

Providing high-quality, personalised experiences at scale is a complex problem, and it stems from the need to discover insights and predict behaviour in real time.

For example, suppose we wanted to provide a customer in the figure above with a personalised product recommendation. We could examine their recent shopping behavior and observe that they are currently browsing the electronics section, specifically camera lenses.

A great example of personalised recommendations in action would be if we could look at the customer's order history to see that they recently purchased a camera at the store. We would then be able to recommend (i.e. cross-sell) the lenses best suited for their make and model.

If we weren’t able to gather that from their shopping history, we could also look at other customer spending histories to find other products they purchased while shopping for lenses (camera batteries, digital frames, straps, bags, mounts etc.) and provide recommendations for these products to our customers, potentially with small discounts to incentivise them to make the purchase.

All of these provide benefits to both the customer and the retail organisation. The customer with the intent to purchase goods in this category receives personalised offers, and the store is able to increase its customer satisfaction and revenue.

However, providing this information in real-time would mean scanning through the tables in the figure above to find related customers and historical purchases and then scanning through products before creating a personalised offer and having that appear as a link on the customer's next page. All within a few seconds! Sure, it’s doable for small storefronts with low amounts of traffic. But what happens when you get hundreds of people browsing every second and scanning millions of purchase records? That’s right, it slows (or even breaks) down!

Customers today have options, and they pick the best one #

To predictably deliver valuable customer experiences at scale, we need to think about the problem a bit differently. We need a more performant and simpler approach to gathering insights from our customers.

A visual graph depicting customers viewing products within a store.

In the figure above, we’ve transformed a subset of our retail data model into a graph data structure. From the model we can see the power of describing entities within our business as entities connected by relationships. The figure shows a customer browsing camera lenses after purchasing a camera. There are a few cases of personalisation depicted in this example:

  • The customer is recommended a camera lens of the same brand as the camera they purchased
  • The customer is recommended products within the same category as the camera they purchased
  • The customer is recommended products associated with other customers showing similar purchase- and browsing behavior

Personalisation and real-time recommendations can improve customer satisfaction and increase revenue for retailers by immediately reacting to consumer intents and by seamlessly providing them with their next best option.

A visual graph depicting a customer being recommended products within a store.

Further benefits to real-time recommendations is the additional value it unlocks for businesses. By being proactive in understanding the intent of consumers as they browse a store, retailers unlock the ability to:

  • Recommend products with a short shelf life and reduce waste
  • Recommend products at risk of becomming dead-stock to customers with intent to purchase within that category
  • Recommend high-margin products
  • Recommend a high-value upsell
  • Unlock the ability to negotiate store-level, and click-level brand deals that convert for customers with intent to purchase

These are a few of the benefits that can come with implementing a graph data management and intelligence platform. If your interested in reading more about what these solutions are capable of, have a look at our writup of knowledge graphs within large enterprises.

Conclusion #

We’ve now seen the power a graph data solution can provide to a customer’s shopping experience. Another benefit of the solution is its flexibility. An extensible data model allows organisations to experiment without much difficulty.

For example, say a retail organisation wants to introduce an innovative sub-15-minute delivery service at a premium price and with a limited catalogue since the items must be packed from a co-located space for speed. This is a substantial change at an operational level, even for large corporates. Changes will have to be made at the store, delivery, customer and, of course, data level. Luckily, we now have a data model that allows us to be flexible. Meaning we could add new customer types, delivery drivers, regions or product types without disrupting the rest of the business.

You don't have to drown in a sea of rigid, complex data. Graph data solutions provide retailers with the ability to ask complex questions and quickly uncover insights that would be difficult to identify using traditional methods. They enable organisations to unlock the ability to ask complex questions and receive insights immediately.

At Rockup, we believe in the power of relationships. If your company can benefit from a deeper understanding of the data that drives your success, contact us to schedule a call.

Subscribe to Rockup’s Insights

We are committed to keeping your data safe.


Too much email?