In his new company, Bull Moose's Brett Wickard uses statistics to infer customer buying patterns

When Bull Moose's computerized ordering system bought hundreds of copies of Walt Disney's “Hocus Pocus,” Chad Verrill, the music chain's director of operations, was surprised. He thought the retailer's customers preferred real horror movies at Halloween. But the computer figured out what store employees didn't: that each October for the last several years, hundreds of copies of “Hocus Pocus,” the 1993 Disney film starring Bette Midler, flew off the shelves.

“The system can overcome our human biases,” says Verrill. “If movies or music were out 10 or more years ago, we might not carry them, but with our special order system at the point of sale, if a couple people order it, the system will bring it back. Otherwise we'd overlook it.”

He figures that rather than not having the movie in stock, the store made a $2 to $3 profit for every copy of “Hocus Pocus” sold. That might not sound like a lot. “But when you only make $2 off something that costs us $10, we don't want to have extra items,” he says. “It's called 'lean retail.'”

Bull Moose's founder and President Brett Wickard, a self-described math geek, embraced the concept of lean manufacturing early in the company's history. Using statistics that projected future orders based on past consumer purchases, he wrote software that assured the 11 Bull Moose stores never ran out of inventory and didn't order items people weren't buying. That software and services supporting it evolved into his new company, FieldStack LLC, which he founded in 2013 in the same former firehouse as Bull Moose's Portland headquarters. FieldStack, at which he also is president, is now selling the lean retail software and services to other retailers, including the Pet Life pet stores. Bull Moose also is a customer, and a test site for new software features.

Wickard says the current product's development started more than a decade ago at Bull Moose. He knew he had something special when someone visited him from what he describes as the biggest distributor in the music business. “He said, 'you have a secret,' because of Bull Moose's inventory metrics,” says Wickard.

“All of their customers had a [product] return average of 15% to 20%, but Bull Moose was in the low single digits,” Wickard says. “We were never out of things and our inventory coverage percentage was much higher. He wanted us to share our secret with the rest of [his] customers.”

Wickard admits Bull Moose did have a secret. “We were looking at the data in a different way,” says Wickard, who majored in chemistry at Bowdoin College. “We are using the people who sell us used goods to infer demand for new goods, that is, we look at whether they sold something back soon after they bought it. We've wired math into our ordering process.”

Verrill at Bull Moose points to another advantage. Because the system can infer — he says accurately — what customers want to buy, fewer employees are needed in the ordering process. Early on, each Bull Moose location ordered its own items to sell. Three to four people were involved in the reordering, Verrill recalls.

“They spent hours buying what they wanted and thousands of dollars on things they thought were cool, but didn't sell,” he recalls. “That could amount to 10 hours a week in 10 stores every week. Now a total of four people do the reordering. And we want the customers to decide on what's being ordered.”

FieldStack's software and services are part of a field known as supply chain management. North American sales of that type of software hit $10 billion in 2014, up more than 12% from the prior year and on a high growth trajectory going forward, according to Gartner Group, a market research firm in Stamford, Conn.

Playing the odds

Wickard says he got the idea for the statistical system, which is based on the probability of customers buying an item in the future, because he loves math and science and sees their value when applied to retailing. One of his heroes is German physicist Werner Heisenberg, known for his uncertainty principle, which relates to how humans affect a particle they're trying to measure. Heisenberg said a precise measurement is not possible, and suggested people rely on probability and mathematical formulations to come up with the best estimate.

“In most things in life, you want to be the least wrong,” says Wickard, who applied the notion of the likelihood a customer will buy a certain item in the future to his software, which he has trademarked.

The software's features evolved from the original Bull Moose program through a series of new companies Wickard founded to advance the product: first was Crickery Wood, then Nanocran and finally FieldStack. Located behind Po' Boys & Pickles in the former Ladder 4 Engine 9 firehouse, FieldStack's lobby is adorned with black-and-white photos of the horse and buggy firefighting days and former pump trucks, while its storage room is stacked with miscellaneous piles of DVDs and CDs from Bull Moose.

The original point-of-sale software used in Bull Moose stores, known as the Willow Retail Suite, factored into an equation the probability that a specific music or movie title would sell. That equation also calculated the likelihood of Bull Moose having that item in stock and at a price that would give the store the best margin. The next company, Crickery Wood, added office applications and an ordering system, and made the software easier to use for the non-programmers at Bull Moose, like Verrill.

“Brett worked on a way for all the rest of us to use it,” says Verrill, who has been with Bull Moose since it started in 1989. “Before that, it was back-end programming” that only a technical person could handle.

Verrill says letting the software run the reordering saves time. “The statistical processes Brett wrote into it do the reordering automatically,” he says. On a typical day before noon the system handles 50 to 75 orders, each with 30 to 150 different items.

FieldStack now has 20 full-time employees, a number Wickard hopes will grow by 50% this year. Revenue in 2014 was in the $2 million to $5 million range, which he expects to rise 30% this year. The company is self-financed.

The company also is selling its software and services to retail, web and wholesale industry clients that handle large volumes of transactions. It installs the software on their point-of-sale computers, and then uses the customer's inventory information to customize the installed software so it can predict and order new products. The software looks at the product type, size, shape, cost-to-shelve, cost-to-store, pricing, demand and other nuances. The system is cloud-based, so FieldStack can offer service to customers without having to be on-site. Customers pay a monthly service fee based on their revenues.

“It is very custom,” says Michael Stefanakos, vice president of strategic partnerships at FieldStack. “We consult with [customers] to spend time before we do the implementation, for flow and resource efficiency. A lot of what we do is help them understand what to buy, how much, how to sell it and the labor involved.”

He said FieldStack helps customers analyze data and steer their business direction to make more money. Since the system is in the cloud, it can automatically check prices at Amazon, eBay and other websites to help price products. Wickard says the addition of an e-commerce component to the software made a dramatic impact on lean retail at Bull Moose and at other customers.

“We help them grow their business,” adds Wickard. “We find a lot of clients are not making money in areas they thought they were. Our system helps them get what their customers are looking for into the store before they [know they] want to buy it.” The software takes into account the geography of the customers and local preferences.

In one case, Wickard says, a client ended up increasing shipping costs, but made more money from other savings, including labor, because the system looks at the overall picture for a customer. “Looking only at expenses is a narrow way to look at things,” he says. Wickard describes the company's lean retail approach as the intersection of resource efficiency and flow efficiency.

Lean retail

The FieldStack system works from a retailer's storage room to the store aisles to most efficiently track and reorder inventory, and even to help identify the best locations to sell items within each store.

“The system reorders based on what people buy,” says Verrill. “Vinyl [record sales] are through the roof in downtown stores in Portland and Portsmouth, N.H. Sales rose two to three times in the last year or two.”

But since vinyl records take up more space than DVDs or CDs, their sales bump is causing Bull Moose to change some store layouts. In Portsmouth, the store is being totally revamped to handle displays of vinyl records.

Sometimes Verrill looks across multiple stores to see why a certain item isn't selling in one store, and then suggests changing its location. He also looks at local trends, for example, the Sanford store sells a lot of video games so more of that inventory goes to that location.

The system works best with used or older items previously sold at Bull Moose. For example, if a famous musician or actor dies, sales of their music or movies typically spike. The system senses the increase in sales and orders more. The same is true if a movie is taken out of print, which is something Disney does occasionally with its cartoon movies to increase their value.

“The system looks at a couple million items each night to see how many people are selling an item and the prices,” says Verrill. “If 100 people are selling it at $1 compared to 10 people selling it at $25, the latter is in the price range where it should be. The system is very good at pricing used items by looking at the last time it came in and how much it sells for right now online in places like eBay and Amazon.com.”

Bull Moose stores average 70,000 to 80,000 SKUs, or stock keeping units, in the nine Maine and two New Hampshire locations. Verrill would not comment on company revenue.

He says one of the biggest savings has been on payroll, by trimming the upwards of four people per store who spent hours choosing and reordering items to four people and the system now.

People are needed to look for mistakes, for example, if the software flags an inventory glitch at a store that says it received one Robin Williams movie, but the computer counts 10 at the store. Another aberration can be individual customers, like a library, suddenly buying 12 copies of an item. The people recognize this is a one-time event and keep the system from reordering a lot of copies based on that one unusual order, Verrill says.

The system also handles individual items better than cases, as it may not know a particular item comes by the case. It's an issue Verrill says FieldStack is addressing now. “Pet Life is a new client of FieldStack, and a lot of their needs are reordering cases [pet food, etc.]. Trading cards, soda and candy also come in cases.”

Competitors include JDA Software Group Inc., which helps clients with retail and supply chain planning and is based in Scottsdale, Ariz., and Manhattan Associates Inc., a supply chain and warehouse-management software company based in Manhattan Beach, Calif.

Wickard says one distinction for FieldStack is it's an all-Maine company, meaning its employees and technology are in the state, and it offers customization and consulting along with its product.

The company does not want to outsource work. It does want to be a technical service provider with its resources in Maine.

“Part of the vision is to build a big software company in Maine,” adds Stefanakos of FieldStack. “This can be successful. We'd like to quit losing our most talented folks out of state.”

Says Wickard, “We want to make it work in Maine. We love it here. Maine has an opportunity to push way forward on technology.”

Read more

Bull Moose to open third New Hampshire store