Forecasting helps Loblaw hit the icing sugar season


By automating its forecasting system retailer Loblaw Companies Ltd. is freeing up its staff to spend more time on the floor with customers, and making sure items those customers are looking for are actually on the shelves.

As Canada’s largest grocery retailer, with brands such as Loblaws, NoFrills and The Real Canadian Superstore, Loblaw has more than 1,200 corporate and franchise stores across the country.

In a presentation at data warehousing vendor Teradata’s annual Partners conference in Orlando this week, Chris Gallagher, vice-president responsible for business systems at Loblaw, told attendees the company wanted to automate the forecasting process and generate more timely data to make better ordering decisions.

He said previously, with short-term, operational-type forecasting, Loblaw was “too slow to respond to peaks and valleys.” The new system, he said, would support long-term forecasting, from sales to supply chain.

Previously, Loblaw relied on historical data and based buying decisions on a four-week rolling average of sales. The problem, said Gallagher, was the system was slow to identify trends. He cited the example of icing sugar. There are typically three to four peak icing sugar seasons in the year, but, Gallagher said, the old forecasting model was too slow to identify them. “If you’re using a four-week moving average, you’re going to miss the uplift and you’re going to overbuy when the season is over.”

According to the Loblaw executive, in addition to better data, automation frees-up staff from feeding raw data into the system and making decisions around what products to order and when.

With Teradata’s Demand Chain Management (DCM) forecasting applications and its own modeling work Gallagher said Loblaw has been able to reduce “out-of-stocks” from 1,700 each week to 300. That means increased sales, as only 56 per cent of the time people buy an alternate product. DCM also helps to reduce the stock of slow-moving items and get rid of dying items for a more efficient supply chain and use of shelf space.

The model sees up to 50,000 items ranked dynamically based on product performance, location, margin and a number of other factors. Items are then profiled and clustered, with a smoothing to allow for casual change in demand. Rates of sale are tracked, going back over multiple time periods to allow for short, mid- and long-range trending.

All of that data is then put together to produce a sales forecast that drives ordering, replenishment and budgeting activities in an automated way.

“We’re not there yet, but we’re trying to get there,” said Gallagher. Looking to the future, he said he’d like to expand DCM to fresh food and perishables. They’ve done a pilot but the challenge is linking purchases, which are by the case, with store sales, which are by the kilo.

Loblaw is also currently piloting the use of Radio Frequency Identification (RFID) to assist tracking. Gallagher said he saw the value of being able to track a product right through the supply chain from the vendor thru the store shelf. There was a lot of work and a lot of infrastructure required to make it happen though, he said.

“If you’re thinking about the RFID business and you’re looking to capture all access point data, you want to put all that into Teradata to be able to crunch the numbers,” said Gallagher. “I think at the end of the day [RFID] will allow you to follow the product, even through the four walls of the store.”


Related Download
Virtualization: For Victory Over IT Complexity Sponsor: HPE
Virtualization: For Victory Over IT Complexity
Download this white paper to learn how to effectively deploy virtualization and create your own high-performance infrastructures
Register Now