Insight: machine learning revolutionising the customer experience
The global business has recently launched a test version of the 'Easy Button' service across its business, including on its it's website and app, that uses the cognitive technology capability of IBM's Watson software to enable customers to verbally make orders, as well as ask questions on various areas such order tracking, and product recommendations.
Speaking at Retail's Big Show (organised by the National Retail Federation) in New York Fasul Masud, CTO of Staples, says: "We can teach Watson how to take unstructured speech and work out the 'intent' of customers and how to then service them. There might be 60 ways of asking for the same thing so we need to enter these 'intents' into the system."
The beauty of machine learning technology such as Watson is that it builds on its knowledge over time so its capabilities increase as it gains experience of dealing with customers' interactions. "It is re-taught any failures it makes. And if it cannot understand what to do then the interaction goes to a live person. We set the thresholds on how confident it is in providing an answer," explains Masud.
At this early stage of implementation he says the solution is simply taking "low class" tasks away from people but he can foresee when it will do a lot more for the Staples business - both internally and for customers: "There are lots of potential use-cases including helping customers' search on the website, acting as the operating system for the e-commerce platform, and Watson could even be answering the [call centre] phones."
Machine learning solutions have been a major feature at NRF this year and Blue Yonder is one such provider. It has been highlighting how it is using the technology to massively improve replenishment for retailers. The company will shortly announce the work it has been doing with a major UK grocery business, which has had a dramatic impact on its performance
By using a multitude of data inputs the solution can identify patterns in previous scenarios and from this make predictions, which can be acted upon by actions such as restocking specific items, changing the pricing, and re-positioning specific goods in-store.
Email this article to a friend
You need to be logged in to use this feature.
Please log in here