omni-channel feature

Optimizing Your Morning Coffee

Imagine this scenario: You open the door to a Starbucks and jump into a line that is six- people deep. Entering through the door behind you is someone on their phone. You see them tapping on the screen and step aside, but not into the line. Before you have the chance to say your order or swipe your credit card, the person on their phone is picking up their coffee and sipping it as they exit. Is this fair?

If you ask queueing theory expert and ISyE Professor Sherwin Doroudi, there is no one-size-fits-all method for businesses to follow when prioritizing in-person and online orders, also known as an omni-channel system. For Starbucks alone, this could raise questions such as, "Which orders do you prioritize first—walk-in or mobile?" and "When should the staff be taking new orders as opposed to fulfilling outstanding ones?"

"It's really contextually driven," Doroudi says. In their recent paper, "Designing Efficient and Equitable Omni-channel Service Systems," Doroudi, ISyE Ph.D. student Kang Kang, and their collaborator at Stony Brook University, Professor Mohammad Delasay, are developing and analyzing omni-channel service models. Their models consider a number of factors, including customer patience levels, demand, workers on hand, and other complexities (see sidebar).

Among their initial findings, Doroudi and Kang have discovered that sometimes a mobile ordering option can be bad for business, even when making the right prioritization decisions. "That was what we found most surprising," Doroudi says. "Allowing users to submit mobile orders on their own should free up staff time, which it does. Yet, due to subtleties uncovered by our research, it can actually lead to more customers opting out of a service for fear of lengthy wait times."

To construct their models, Doroudi and Kang chose coffee shops as their archetypical example. Their reason was simple: most modern coffee shop orders filter in through two streams—online or in-person. Developing stochastic and optimization models in this environment would be much simpler than a healthcare clinic or government agency, for example, which typically feature multi-step processes.

But for Doroudi and his collaborators, there's far more than coffee at stake.

"We don't just want to make things as best as possible for the business, the service provider, the mobile crowd, or the walk-in crowd," Doroudi says. "We want to look at all stakeholders and see what are the trade-offs."

In their research, Doroudi, Kang, and Delasay determined there is the potential for unintended discriminatory treatment against low income, disabled, and other individuals who are less likely to have access to smartphones or cannot use a mobile app. Ensuring equitable outcomes will require paying careful attention to how each stakeholder is treated under a given operational recommendation.

While saving or losing a couple of minutes at the coffee shop every now and then may not mean much for most people in the grand scheme of things, the impact on waiting times is much more critical in healthcare and government systems, which feature similar (albeit more complicated) system dynamics. In healthcare, for example, newly emerging mobile health and self-diagnostic technologies are beginning to allow some patients to bypass lines and procedures at the clinic, thus freeing up provider resources. But these technologies will (at least initially) be available primarily to those from the wealthiest backgrounds. How can the potential benefits of these technologies be balanced with the additional delays they impose on those who are less fortunate? How (and to what extent) can those additional delays be mitigated? Expanding the scope of their omni-channel models to answer these kinds of questions in the context of hospitals, government agencies, and nonprofits, is precisely where Doroudi and Kang plan to turn next.

"We've been developing new tools and new ways of looking at things," says Doroudi. "Once we have the full picture, we can show people a graph or a formula to tell someone if it's better to offer omni-channel at all and whether walk-ins should be prioritized. These new ways of looking at these systems we're developing could actually be one of the important findings of our research."