More accurate data and streamlined systems give L.L.Bean an omnichannel edge


What’s a catalog-dominated business to do when online and other digital channels invade the world? If you’re L.L. Bean, you evolve.

“When you’re in the catalog business and you have control over which products you’re putting in front of customers and when, you have a higher degree of predictability,” says Kirsten Piacentini, vice president of inventory management for the 105-year-old retailer headquartered in Freeport, Maine.

Though still subject to forecasting error, retailers with a catalog-heavy presence have historically enjoyed significant control over their marketplace. But that’s all changing.

From push to pull

In today’s pull environment — where consumers can decide which products they want to look at online and respond to ads on a variety of channels including TV, social media and other digital platforms — retailers need new tools to navigate the customer-demand waters.

“It’s a lot harder to predict to what and when they’re going to react, and in what quantities,” Piacentini says. Though L.L.Bean has deep historical data and internal institutional knowledge, the team still found it difficult to manage the changes they were seeing in product demand.

A conglomeration of legacy platforms supporting the company’s forecasting and inventory functions wasn’t making demand predictions any easier. “We had multiple systems trying to forecast,” Piacentini says.
The team was also using forecasting processes built for the catalog model, which proved ineffective for ecommerce: Monthly online forecasts didn’t react fast enough to the weekly trend cycle.

“For retail, we didn’t have good visibility on store-level inventory when making buys for the next season,” Piacentini says.

Changing how forecasting and inventory purchasing is handled for a company the size of L.L.Bean — and with a staff of only 70 people to do it — doesn’t happen overnight.

The combination of drivers moved the team to implement JDA Software’s demand and fulfillment platforms to improve forecasting accuracy and increase visibility. Demand forecasting fragmentation is a problem for a growing number of retailers; multiple systems are often melded together with varying integration success.

“Applications have been developed in silos,” says Jim Prewitt, vice president of North American retail strategy for JDA Software. Internal sub-groups have launched their own solutions over the years, but as the number of channels has grown to include catalog, phone, in-store and ecommerce, retailers have found the patchwork of platforms can’t keep up.

“Those organizations within their silos have made decisions to purchase applications and use them, but they haven’t shared that information across the organization,” Prewitt says. Today’s more complex forecasting needs call for comprehensive and timely forecast data, as well as the ability to use that information across the enterprise.

Changes in consumer preferences and habits compound the problems retailers encounter when using disparate forecasting systems. Creating a forecast that provides actionable insight from raw goods to finished products in the distribution center is increasingly important for businesses navigating today’s retail environment.

“They need to use their demand forecasts to help with their buy for merchandise and their execution, but also upstream all the way into sourcing with contract manufacturers across the different levels of their supply chain,” says Keith Adams, JDA Software’s vice president of sales.

Designed for results

Piacentini, who has 14 years as an IT project manager under her belt, won’t sugarcoat the amount of effort the company put into the deployment of the JDA platform. “It was a lot of work,” she says. “But it was so worth it.”
She knew it was important to begin with a strategic, long-term point of view. A strong commitment to organizational change management also allowed the team to stick with the implementation through its multi-year time period.

“The whole program for business transformation is a seven-year program,” Piacentini says. The forecasting deployment took place during the first three years of that window, having launched in the fall of 2014.

“It took probably the first two years post-implementation to really sort everything out and feel like we were starting to see significant benefits from it,” she says. It sounds like a long time, but changing how forecasting and inventory purchasing is handled for a company the size of L.L.Bean — and with a staff of only 70 people to do it — doesn’t happen overnight.

“You need time to design, build, test, implement,” she says. An internal group was dedicated to the post-implementation plan, enabling the team to continuously improve processes and maximize use of the new system.
Customer service levels in the L.L. Bean retail channel have improved with the launch of JDA. “When ordering product from vendors for our retail stores, before we only had chain-level data about how much inventory we had,” Piacentini says. “We didn’t have that information at the store level.”

With about 30 stores, even small inaccuracies could result in under- or over-buying, quickly impacting product availability. “Now we have that store-level detail, and our customer service levels have seen 10 percentage points’ improvements as result,” Piacentini says.

The length of the team’s forecasting abilities has also increased, while the time needed to update those forecasts has gone down. “We’ve gone from forecasting over a much shorter time horizon to now looking over the next 18 months,” Piacentini says.

Rather than continuing to manually develop forecasts, the new system downloads data for every product in every channel automatically each night. “It improves your decision-making because you can more quickly identify winners and losers,” Piacentini says. The team can work with vendors to chase popular products, or evaluate promotional and markdown opportunities for items that aren’t moving as well.

Forecasting accuracy has also improved. “We now measure forecast accuracy at a much lower level of detail,” Piacentini says. The new granularity encompasses everything from item to color to channel. Since going live with the system, the company’s accuracy has improved by four percentage points.

Real-time reactions

A drive to change how forecasting results are measured is affecting retailers across the spectrum. Forecasting needs have become more complex, Prewitt says: “A simple moving average might have been fine in the past.”
Remaining competitive in today’s business environment, however, means having the ability to very quickly recognize slow-moving items, evaluate seasonal demand earlier, spot fast-trend items and take advantage of products with short lifecycles.

“It’s much more important to be able to react to those very specific items,” Prewitt says. Retailers increasingly need forecasting built around those factors instead of more simplistic algorithms that may have met their needs in the past.

Other advantages, some less measurable but no less impactful, have become apparent in the form of what Piacentini calls “soft metrics.” The team was previously juggling six forecasting tools but now needs just one to accomplish everything.

“One of the benefits is increased visibility across the entire company on what the forecast is,” Piacentini says. Different systems taking in and churning out potentially different data can hinder the process significantly. “They might all be right, but they’re telling different stories,” Piacentini says. “Now there’s one version of the truth.”

As L.L.Bean’s various internal groups talk with their colleagues — from marketing staff to merchants — everyone can use the same forecast to underpin their efforts. “It also gives vendors visibility into what our forecast is over a longer time horizon,” Piacentini says.

Suppliers use the data to plan production capability in their own facilities, to make commitments on fabric colors and to determine when and in what quantities raw materials should be ordered.

“They now have that information and can do something with it,” Piacentini says. “We’ve cut lead times as a result.”

Julie Knudson is a freelance business writer who focuses on retail, hospitality and technology.


Comments are closed.