On Your Mark, Headset, Go!
To increase fulfillment efficiency and order picking accuracy, eBay Enterprise gives its warehouse a voice.
Warehouse workers at eBay Enterprise, a King of Prussia, Pa.-based retail order management and fulfillment provider, used to employ radio frequency (RF) handheld units for picking and packing orders, but found them counterproductive. With help from a voice-enabled technology solutions provider and data collection hardware manufacturer, eBay Enterprise implemented a hands-free, voice-based system to increase order accuracy and improve fulfillment efficiency to shorten delivery times.
eBay Enterprise provides distribution, warehouse management, inbound and outbound logistics, and returns processing from seven campuses across the United States, Canada, and Europe, and serves more than 60 countries around the world.
"The fulfillment division provides order fulfillment for both brick-and-mortar and e-commerce retailers," says Jason Adamchik, director of engineering for eBay Enterprise. "To do this, we must make sure we have the right technology and warehouse capacity, and that new technology we acquire delivers performance productivity and accuracy improvements."
For many years, eBay Enterprise used RF technology to assist with picking and packing orders in the warehouse. But employees had to hold the RF units when they needed to operate in a hands-free mode.
The Choice of Voice
"We began to analyze whether it made sense to supplement our RF technology with voice, which would give our employees more freedom to be hands-free as they worked," says Adamchik. "We thought our warehouse facility in Louisville, Ky., would be a logical place to try voice-based technology. The Louisville facility serves many of our highest volume, highest unit clients. If we switched to a voice-based order picking system, we would eliminate the need to handle RF devices, which would improve productivity and accuracy."
It wasn't just high-volume order fulfillment challenges that Adamchik and his staff was trying to solve. There was also the issue of handling enormous seasonal spikes in order volumes, which the warehouse addressed by hiring temporary workers to augment regular crews.
"A major challenge was getting these temporary workers up to speed fast, so we were continuously training," says Adamchik. "We thought we could reduce training time and error rates by moving to a voice-based system."
The move to voice-based picking did exactly that. Across the board, warehouse order-picking error rates dropped 25 percent—including the temporary workers' performance. For the temporary warehouse workers themselves, training times with voice-based picking were cut in half.
At Your Command
"Warehouse workers wear two elements of the voice-based system," explains Adamchik. "The first piece of equipment is a headset equipped with a microphone. The second piece is a voice-based unit workers wear on their belts."
eBay Enterprise works with each new employee to set up a voice "template" that is specific to the characteristics of that employee's voice. This includes training the system so it can readily recognize each employee's voice inflections or accents. To facilitate the learning process, the new employee is asked to recite a phonetic alphabet in either English or Spanish while the voice technology listens and detects unique characteristics of the employee's voice.
Once voice training is complete, the employee learns how to handle the various prompts that the voice-based system provides throughout the order fulfillment process. To use the voice-based technology, employees have to master the picking process so they know exactly how to answer the prompts and interact with the system.
"To fully engage with the voice-based system, employees have to understand the warehouse operation," says Adamchik. "They must gain thorough knowledge of the process so they know when to place totes on carts, how to pick orders, when to load, and how to pack."
eBay Enterprise uses a voice system that consists of a Honeywell Vocollect Talkman A730 voice unit that is attached to the employee's belt, an SRX2 wireless headset, and Vocollect Voice Catalyst software. "Workers wear ruggedized devices integrated with a central software that screens out extraneous noise in the background," says Jay Blinderman, director of product marketing for Vocollect by Honeywell.
This voice-based system assists the employee through the warehouse picking process, offering simplified prompts such as next, ready, repeat, go to 213, or pick three. By keeping voice commands simple and short, the voice-based system is able to accurately "hear" what the worker is saying. A second benefit of the abbreviated voice commands is that the system can easily translate them into a digitalized piece of information that can be sent to a warehouse management system (WMS) for integrated and automated order tracking.
"Before selecting a voice-based order picking system for the warehouse, we did our due diligence," recalls Adamchik. "Internally, we felt the technology made sense, so we talked with several different solutions providers before we ultimately made the decision to go with Mountain Leverage as our system integrator and solution provider, and Vocollect by Honeywell for our hardware and software. The Mountain Leverage solution was appealing because of its flexibility."
"The business model at eBay Enterprise's warehouse was variable, given seasonal sales spikes and the need to dramatically increase the workforce during those times," says Alex Reneman, president of Mountain Leverage, based in Grafton, W. Va. "eBay Enterprise needed a flexible and scalable solution, due to that variability. From both deployment and pricing perspectives, the model had to scale both upward and downward.
"We designed a baseline subscription plan with incremental rental provisions for extra equipment as needed," Reneman says. "That decreased equipment rentals at times of the year when eBay Enterprise didn't need the extra capability."
Making Changes on the Fly
Despite the advance planning and due diligence, eBay Enterprise encountered several wrinkles as it rolled out the initial implementation of warehouse voice.
"In any implementation, no matter how thoroughly you think you have planned, you have to make adjustments," says Adamchik. "We saw several instances during deployment where it was advantageous to make some changes. And we found that we could make those changes on the fly."
The ability to modify the technology quickly was also encouraging because of the volatility of eBay Enterprise's seasonal workforce. "Any plan we devised had to address workforce fluctuations," says Adamchik. "We run a 24/7 operation. We need to ensure we always have enough headsets available, as well as extra batteries so operations continue to flow without interruption."
A second area of focus during implementation was integrating voice-generated commands into the WMS. "We needed to understand all the feeds into the WMS and be sure we integrated them all," says Adamchik. "A second and related phase was whether our network had sufficient capacity to add new devices that would transmit to the WMS. This required some network revisions to accommodate the additional communications, and ensure adequate network capacity."
Much of the systems integration work was eased because the company was able to use industry standard application programming interfaces to get systems talking to each other. For operational integration, there was not much change on the packing side of warehouse operations, but some procedural adjustments had to be made on the picking side. These changes primarily involved enhancing existing operational procedures.
A third and critical area of implementation concerned the operational revisions that voice-based technology required.
"We needed a strong change management process to introduce this system without encountering any hiccups that would disrupt our production schedule," says Adamchik. "We took our time and laid out all the specifications for replacing RF with voice."
Because the move to voice impacted warehouse processes, eBay Enterprise immediately involved its operations group in the project team, and used their input as the basis for making operational changes.
"We had some pickers who had been with us for a long time," says Adamchik. "They were used to the RF system, and were initially not excited about changing how they did their work."
To ease adoption, eBay Enterprise took advantage of the fact that its warehouse operates in 12 separate work zones. "We phased in the voice-based technology by zones," says Adamchik. Breaking the project into individual implementations by zone allowed project team members to distribute the new technology incrementally, and to ensure it was embedded and running smoothly in one zone before the team moved on. This approach gave eBay Enterprise more control over troubleshooting technology and operational workflow issues, as well as ensuring that employees were being trained and acclimating to the new processes.
"Incremental implementation was well received," notes Adamchik. "Employee feedback was positive, and most employees now say the voice-based picking and packing system makes their jobs easier."
Going forward, operations, engineering, and pickers are "continuing to look at new enhancements and tweaks in the warehouse," Adamchik says. "We are at a point where we feel comfortable with the new system. We are also happy with its ability to quickly make small modifications. Now we are looking at areas throughout the company where we can deploy the technology in the future."
Adamchik is certain that eBay Enterprise will achieve a rapid return on investment. "We took the time to lay out all our requirements first, with everyone across the organization—operations, systems, and management—on board," he says. "We also wanted to be certain that the benefits of the system would cover our initial investment, so we spent some time on ROI modeling. Finally, we vetted multiple vendors that offered voice-based picking solutions.
"Our goal was not only to find the right technology for what we wanted to accomplish, but also to find a capable business partner who would stay with us for the long haul. In the end, what made our project a success was a committed and multidisciplinary work team where systems, operations, and engineering personnel all pulled together."