Site navigation

M&S to Replace Call Centre Staff with Chatbot AI

Theo Priestley


virtual assistant chatbot

Marks & Spencer is replacing 100 call centre staff with ‘chatbot’ artificial intelligence designed to quickly deal with phone call customer complaints.

Hot on the heels of the recent announcement that the retail chain is creating an academy to offer data science training to staff, Marks and Spencer are to replace 100 call centre employees with chatbot technology designed to answer customer complaints more efficiently.

The company is said to be using software from software providers Twilio and Google to answer queries and route calls faster. Previously the human operators would transfer callers to the right departments, but Twilio claims the new technology correctly identifies 90% of queries and can direct a call within seconds.

The new technology will be used in all 640 M&S UK stores by the end of September, as well as its 13 UK call centres. No jobs have been lost in the change according to the company, and the 100 displaced staff will now be reassigned to in-store customer facing roles – although bizarrely the retailer plans to shut over 100 retail outlets by 2022 meaning those jobs may not be as safe as they seem.

Chris McGrath, IT programme manager at M&S, said “We know that M&S needs to modernise.

“Twilio’s flexible cloud communications platform has enabled Marks & Spencer to experiment like a startup, while executing like an enterprise.

“We were able to prototype a solution in just four weeks and put it to the test during our busiest retail days of the year. The new solution has given Marks & Spencer an improved ability to have more direct and meaningful conversations with our customers, which also helps us reallocate valuable staff time. We’re excited to see where the platform takes us as we continue the roll out across our contact centres.”

Customer Communication for a Digital Age

Prior to implementing the new chatbot solution Marks & Spencer’s infrastructure was built on a mixture of legacy phone systems that could not support its digital strategy going forward.  As a result, M&S was unable to centralise customer information and could not seamlessly connect customers across its stores nationwide.

After a successful trial with Twilio, Marks & Spencer, in collaboration with DVELP, deployed its intelligent natural language routing solution nationwide and is able to;

  • Handle more than one million inbound telephone calls per month.
  • Transcribe the customer’s speech into text – by leveraging Twilio’s Speech Recognition tool, Marks & Spencer can analyse the voice of the customer in real time.
  • Determine caller intent – through integration with Google DialogFlow, Marks & Spencer is able to take the transcribed text and determine why a customer is calling.
  • Route calls – IVR uses caller intent to route the call to the appropriate department, store or contact centre agent to resolve the customer inquiry.

“Today’s consumers have come to expect a great customer experience from the companies they buy from and communications are increasingly central to this” said Rob Brazier, director of product management at Twilio.

“We are thrilled to be working with such an iconic retailer as it transforms its communications in order to deliver the best possible experience for its 32 million customers worldwide.”

Death of the Call Centre?

According to a survey conducted in 2017 by Juniper Research, the use of virtual customer assistants like chatbots will jump 1000% by 2020. However 75% of consumers polled believe that a chatbot would not be efficient for more complex complaint handling and still prefer to talk to a human.

Gartner predicts that 85% of all customer service interactions by 2020 will be managed by chatbots rather than humans, and another survey identified that 80% of call centre queries cover the same 20 questions at most, highlighting the impending disruption on the call centre industry and that the move by M&S may well be a sign of things to come.

Theo Priestley

Latest News

%d bloggers like this: