College bookstore giant Follett enhances its online experience with AI


Quite a few retailers would like to have the “problems” that Follett faces. The privately owned, suburban Chicago corporation traces its roots back to the 1870s and has become the premier operator of college bookstores in North America. More than 1,200 Follett outlets are spread across campuses and the company operates another 1,800 ecommerce sites selling textbooks and merchandise.

With built-in customer bases, long-term contracts with institutions and the growth of higher education, what could go wrong? Just about anything.

“We don’t have the advantage of selling products through one branded outlet,” says Roe McFarlane, Follett’s chief digital product and marketing officer. “We’re operating in more than a thousand hyper-local markets. Notre Dame merchandise is not going sell on the sites or in the stores of other schools and vice versa. We’ve got a broad customer base that is looking for a narrow range of products.”

This means despite having 1,200 bricks-and-mortar stores, a customer can’t buy a shirt at Follett’s “State U” store and return it at the “Private U” store across town because of the different supply chain. Each store and site operates like an individually owned outlet, even on campuses. At some large schools, stores located near graduate school buildings carry different products than stores near freshman housing.

‘Headless commerce’

Although most of Follett’s merchandise buying capabilities are automated, human oversight is still needed to make sure Northeastern Illinois University’s bookstore sites don’t sell Northwestern University sweatshirts. Getting that extra oversight was one of McFarlane’s goals.

Follett has used IBM for its ecommerce infrastructure for nearly 20 years; McFarlane credits IBM with helping adjust to the mobile-first preferences of customers. Then a study he commissioned last year found some gaps in the company’s ecommerce strategy. It looked at how Follett was projected to continue to scale up, with nearly 2,000 sites, hundreds of thousands of unique apparel items and millions of products.

The company also had to deal with a few big rushes each year, as students and their families buy books and supplies at the beginning of each semester and quarter. About 70 percent of these purchases are buy online, pick up in store. McFarlane saw that eventually Follett’s infrastructure wouldn’t be able to handle the traffic.

“We talked to our partners at IBM and discussed what we found to see what they could do,” says McFarlane. “They’ve been with us a long time so they know there aren’t many businesses like us.”

After more meetings and tests, Follett agreed to invest $50 million last year to incorporate IBM’s WebSphere V9 into its platform. The official rollout will be next spring, and tests of the system show promise, namely the company’s movement into the burgeoning world of “headless ecommerce.”

“In headed commerce, as customers are browsing, shopping and checking out they’re pinging the whole stack, the application, servers and databases,” McFarlane says. “In a headless platform, there are micro-services at the very top and you’re calling on the various services as you need them. It’s much more efficient and is where heavily trafficked sites are headed.”

Newer platforms such as WebSphere V9 rely more heavily on algorithms and artificial intelligence to streamline operations. “When you have thousands of unique sites like Follett, it can be difficult to address customer needs that spread across these different locations,” says Robert Hearn, general manager and global chief revenue officer for IBM’s Watson Customer Engagement. “Only 3 to 10 percent of visitors to an ecommerce site convert to sales, so any minute thing that can be done to make the experience better will make a huge difference.”

“Younger consumers don’t monopolize mobile use. Most people are surprised that our primary searchers are moms shopping for their kids. … We’re catering to more than young adults with backpacks.”
— Roe McFarlane, Follett

One aspect of the new Follett platform will be a boost in struggle detection. Using IBM’s Watson AI, the system will be able to follow customer paths around the various sites and see potential sticking points that prevent a sale.

“The system can then suggest solutions to make that conversion based on past experience,” Hearn says.

AI also will be able to offer better product suggestions when customers are shopping. “It will know from the paths other shoppers have taken what they may be interested in,” McFarlane says. “An art student looking for a particular brush may want to see a wide selection of other brushes purchased by similar buyers.”

Additional services

Follett’s new platform will be centered around mobile users, but not because of its large millennial customer base. “Younger consumers don’t monopolize mobile use,” McFarlane says. “Most people are surprised that our primary searchers are moms shopping for their kids. Also about 60 percent are non-traditional students. These include people returning to school after working or being in the military, single parents and retired people who are looking to enrich themselves. We’re catering to more than young adults with backpacks.”

The other requirement that Follett was looking for with its new platform was an easy ability to expand into services. “People are shopping for books and merchandise, but we wanted to explore business ‘adjacencies’ that we could enter,” McFarlane says. “There’s no reason buyers won’t be able to book tutoring through our sites, as well as travel and look for jobs and internships. We also want to continue to follow the trend toward more digital course materials as well as open content.”

Shifting into services means making some fundamental changes to how Follett’s sites work. “When you buy a hoodie, it gets moved into your shopping cart and you purchase it,” McFarlane says. “It’s simple and we’ve done it millions of times. With services, however, you need to hook into the application program interfaces of other companies. For instance, you want to allow a tutoring company to offer their services through your site, helping them find customers that come through our front door on the web. We needed to have an advanced system that would take us to that level.”

A more robust AI program is also expected to make each school site even more personal. Current protocol when rain is forecasted in an area is for the Follett sites to feature rain gear on the front pages. It has to be done manually by a tech now; with the new IBM system, local changes will be done automatically. AI will also provide an extra pair of eyes to make sure products offered are timely and on target.

One future area where McFarlane sees great potential is using the system to expand Follett’s customer base. “Presently our customers are students and their immediate families, but there’s no reason why we can’t do what other retailers have done in the wedding business: Create a registry or ‘wish list’ of products that students can share with grandparents and extended family. Overall, adding AI will bring us new opportunities we haven’t even thought of yet.”

John Morell is a Los Angeles-based writer who has covered retail and business topics for a number of publications around the world.


Comments are closed.