AI Case Study
Sears Appliance Repair Unit gains 5 - 6% productivity by predicting parts required for field worker repairs
Sears built a predictive tool to ensure that appropriate parts were available for home repair visits enabling it to optimise its network operations and improve productivity in the workforce.
Consumer Goods And Services
Personal And Household Goods
"The company developed 'cases,' or repair scenarios, each accompanied by a set of questions a call-centre operator would pose to the person calling about a malfunctioning appliance. It used the data from its 8 million calls per year to predict that, say, '95% of the time when somebody calls about a particular item, this is the part that's needed.' The final key was to then send that part to the person’s home in advance of the service call.
Sears took out 450,000 unproductive calls out of a total of 8M annually. This resulted in the ability to grow the business by 5% to 6% without any additional costs.
"The division hasn’t yet eliminated that many calls but expects to based on its data analysis."
"It used the data from its 8 million calls per year to predict that, say, '95% of the time when somebody calls about a particular item, this is the part that's needed.'"
"As a stand-alone company, Sears Home Appliances and Services would be comfortably within the Fortune 500 itself, with revenue of $8 billion."
A key challenge was to reduce costs and productivity in their home service operation which "performs 8 million of the 22 million appliance repairs that take place annually in the United States.
Sears’ appliance-repair trucks carry about 400 parts...which is sufficient to complete 70% of repairs on the first visit to a home. The 30% of service requests that take multiple visits to complete may sound like a lot, but...it’s an industry-wide issue. Even to reach a 75% first-time success rate would require carrying 8,000 parts...and 175,000 parts would be needed to complete all repairs the first time out.'
'The way you grow a service business is to free up capacity and not run unproductive calls, saving that capacity to run the next productive call.'"
8 million calls per year were analysed.