Propulse Analytics Resources
White paper #1 – Short Form
Deep Thought: Propulse Analytics Is Using Deep Learning To Reinvent Online Shopping
Let us start by stating that U.S. E-commerce has officially entered a mature stage. We’re not saying this because digital sales have plateaued. We’re saying this because a progressively smaller group of new shoppers is now entering this market every year despite a healthy growth in overall sales (see graph A). This, of course, has significant implications on the strategy individual businesses will have to adopt in order to ensure success and in some cases simply survive: Efficient shopper to buyer conversion and repeat selling are no longer important, they are essential.
Among the many technologies and tools that can help E-tailers, we believe the most compelling ones involve artificial intelligence (AI). The evolution AI offers is so important that it simply cannot be ignored. The greatest leap is that provided by the technology called deep learning.
The value E-tailers offer their customers boils down to how efficient they are at matching product and customer needs and doing so with one’s own brand experience. This implies that in order to secure conversion and recurrence, being good at presenting the right product, to each customer at the right time becomes one, if not, the key differentiator. Among options that exist today, machine learning (ML) based product recommendation engines have taken the lead. They generate recommendations tied to a product once it’s selected by a shopper (i.e. “people who selected this also bought that”). A good step forward that is now about a decade old. However, in an era where competitors must use elbows to make room on the court, a more impactful approach is offered by new technologies.
Propulse Analytics has been working on such a solution. We have focused on what we believe is the core issue: Understanding customer tastes in real-time. Tastes change with each session and with a customer’s spontaneous desires. If a company was able to see what customers see (and want) through their eyes as sessions unfold, we believe it would be a lot easier to offer what they want, when they want it. Propulse has done just that and has recently completed consumer tests to observe the impact of this new technology in a live environment.
Without leveraging any personal data, product details or transaction history – only images – we built a mock E-commerce shop and exposed anonymous consumers to it. Behind the scene, the taste recognition engine was running. The results confirmed the initial thesis but also came with a surprise.
Within 1 minute of entering the site, 90% of users decided to forego the traditional product categorization in favor of the engine’s recommendations. In fact, despite 100% of shoppers stating they want recommendations and only 10% having ever purchased from the latter, all forgot that the engine was recommending and started browsing almost exclusively using the engine’s recommendations within 90 seconds. The surprise came from the fact they did so with little to no consideration for price until the very end of their shopping sequence.
What this test proved is that by uncovering and utilizing a user’s unique and real-time tastes, you can engage shoppers to such a level that the online experience starts to actually closely mirror the personalized experience of the retail setting. The experience essentially transcends the digital barriers.
Given the increased importance of repeat customers and brand loyalty in the coming years, it is becoming clear that to be successful in the new age of e-commerce, one has to make shopping engagement a central focus of the strategy. Propulse Analytics believes that early movers towards these remarkable new (and soon to be available) AI technologies will become the dominant forces in the industry.
Note: This is a condensed version of the complete paper.
Our full white paper on the future of E-commerce recommendations will be available for download here. Check back shortly.
Get in touch for more information.