ROI Case File No.496 'The Day the Switchboard Released the Front Desk'
The Day the Switchboard Released the Front Desk
Chapter One: The Phone, and the Guest in Front of You
"Half of front desk work is answering phones."
Mitsuteru Iwase, operations director at GlobalStay, opened the front desk work records. External lines ring during check-in handling, internal calls come from guest rooms, accommodation reservation inquiries arrive—there are time slots when multiple phones ring simultaneously. "Service to the guest in front of us is interrupted by phone calls. This continues."
"Three part-time staff handle phone responses?" Claude confirmed.
"Yes," Iwase answered. "The three concurrently handle front desk and switchboard duties. The old switchboard hasn't been updated, and routing of multiple lines is manual. Six external lines, plus internal lines. Whoever can pick up when it rings, picks up."
"Have you received past solution proposals?" I asked.
"Multiple," Iwase answered. "However, many proposals were bundled with switchboard equipment updates, and the capital investment was large. Since group tourism from China declined, securing domestic sales has been the priority, and we couldn't commit to large capital investment."
"You're also concerned about whether part-time staff can use it after implementation?" Gemini confirmed.
"I am," Iwase replied. "Many have little experience with IT tool implementation. Implementation without usage is meaningless. If there's a way to measure adoption with indicators, we can proceed with confidence."
"Implementation effects are measured by utilization rate and adoption," I said. "HEART suits that design."
Chapter Two: HEART Asks About Five Adoption Indicators
"This case requires HEART."
Claude wrote five letters on the whiteboard: H, E, A, R, T.
"HEART is a framework that measures user experience through five indicators: Happiness, Engagement, Adoption, Retention, and Task Success," I explained. "It's a framework Google developed, used in UX evaluation, but it's also effective for adoption evaluation of internal tool implementation. Not just implementation, but designing through to using continuously and producing results. With tools like AI operators that the field uses, design through to adoption determines success or failure."
"Let's first measure current costs," Gemini said, opening ROI Polygraph. She entered the work records from Iwase.
"Monthly phone duty costs are out," Gemini read. "Three part-time staff spend 300 hours per month on phone response. At 1,700 yen per hour, that's 510,000 yen monthly. Customer service quality decline and opportunity loss from front desk interruption estimated at 300,000 yen monthly. Additional night shift coverage for late-night and early-morning calls: 200,000 yen monthly. Reservation opportunity loss from missed calls: 120,000 yen. Total: 1,130,000 yen monthly. Annualized: approximately 13.5 million yen."
Iwase studied the figures. "Cost from interruption is larger than expected."
"Now let's design with HEART," I continued.
[H—Happiness: Place Part-time Staff Satisfaction at the Center]
"First, we set field satisfaction improvement as the goal," Claude said. "The state where the three part-time staff are released from phone response stress and can focus on the guest in front of them. This is the substance of satisfaction. Conduct satisfaction surveys quarterly after implementation, tracking quantitatively."
"Setting satisfaction improvement as the goal, not just workload reduction," Iwase confirmed.
"Yes," I responded. "Without satisfaction rising, it doesn't take root. If the field thinks 'I don't want to use the AI operator,' it eventually reverts. By setting satisfaction as the first indicator, the priority of selection changes."
[E—Engagement: The Depth of Use Cases]
"Next, we measure engagement," Gemini continued. "Beyond just answering phones with the AI operator, engage it in multiple duties—reservation changes, inquiry responses, in-house guidance. Engagement indicators are the breadth of work types AI processes and daily AI processing volume."
"Implement single-function first, then expand gradually," Iwase said.
"Yes," Claude agreed. "Suddenly delegating all duties to AI causes field resistance. First narrow to first-line response on external lines only, expand to reservation changes and in-house guidance after acclimation. Design that gradually deepens engagement."
[A—Adoption: Lower the Implementation Barrier]
"Adoption rate is the indicator of whether the field starts using it," I continued. "Rather than completely replacing the existing switchboard, we use a parallel connection method for the AI operator. Phone numbers don't change. Existing familiar duties remain, and only AI-handled duties are carved out. Implementation anxiety drops."
"Capital investment can be suppressed?" Iwase confirmed.
"We use a cloud-based AI operator service," Gemini answered. "Operations begin with just phone number forwarding configuration. Initial investment drops by an order of magnitude. This is the Adoption design."
[R—Retention: Mechanisms That Continue to Be Used]
"Retention is tracked at three months, six months, and one year after launch," Claude continued. "Tools that don't take root see utilization drop within three months of launch start. Track AI processing volume and field satisfaction monthly, intervening early if there are signs of utilization dropping."
[T—Task Success: Practical Success Rate]
"Finally, task success rate," I continued. "When AI answered the phone, how much did it solve the customer's request. Cases not solved escalate to humans. Measure success rate monthly, maintaining 90 percent or higher. If it drops, improve AI response patterns."
[Calculating Investment Recovery]
"Let's run the numbers with ROI Proposal Generator," Gemini suggested.
- Initial cost: AI operator contract, phone number forwarding setup, response scenario creation, field training. Total: 1.2 million yen
- Monthly cost: AI operator usage fee 180,000 yen
- Monthly reduction: Phone response workload reduction = 280,000 yen (180 of 300 hours processed by AI), customer service quality improvement from elimination of front desk interruption = 200,000 yen, night shift reduction = 150,000 yen, missed call improvement = 80,000 yen. Total: 710,000 yen monthly
- Net monthly reduction: 710,000 − 180,000 = 530,000 yen
- Payback period: 1,200,000 ÷ 530,000 = approximately 2.3 months
"Payback in just over two months," Gemini summarized. "The reason for short-term payback is suppressed capital investment. By choosing cloud-based, the wall that had been blocking consideration was removed."
Iwase confirmed the figures. "When receiving past proposals for full equipment renewal, initial investment was the wall. With parallel connection, that wall doesn't exist."
"Designs that lower implementation barriers are the first Adoption in HEART," I responded.
Chapter Three: Building Usage Together with the Field
"Let me organize the approach," I said, standing at the whiteboard.
"Weeks 1–2—Compare three AI operators, sign trial contract. Week 3—Design response scenarios, extract common inquiry patterns. Week 4—Phone number forwarding setup, train three part-time field staff. Week 5—Launch with limited operation for first-line external response only. Weeks 6–12—Monitor operations, continuously refine response scenarios. Week 13 onward—Expand functionality to reservation changes and in-house guidance."
"What if the field gets stuck?" Iwase confirmed.
"We provide a support desk," Claude answered. "For the first month, set up a structure where vendor-side support can directly contact the field. Questions that didn't surface during training will inevitably emerge after launch. Whether or not there's a window where you get answers in 30 minutes when they emerge changes the probability of adoption."
Iwase took notes and said, "The viewpoint of designing through to adoption has been weak."
Chapter Four: The Day the Front Desk Could Look Only at Guests
Seven months later, a report arrived from Iwase.
Three months after AI operator launch, 72 percent of external calls completed via AI. The remaining 28 percent escalated to humans, mainly cases requiring handling of complex content or emotional customer responses. Task success rate was maintained at 93 percent, and Task Success exceeded its target among the HEART indicators.
The biggest change appeared in front desk work quality. Phone interruptions during check-in handling decreased from an average of 12 times per day to 2. "Time for focusing on the guest in front of us increased," part-time staff comments arrived. In quarterly Happiness indicator surveys, work satisfaction rose 35 points compared to before launch.
Reservation opportunity loss also improved. Cases of missing late-night and early-morning calls became structurally zero, and reservations in those time slots increased by an average of 12 per month. Compared to before, when late-night inquiries were pushed to next-day handling, more cases now completed reservations on the spot. "AI not sleeping has become a value to customers," Iwase wrote.
Functionality expansion proceeded smoothly. From week 13 reservation changes, from week 16 in-house guidance—AI processing scope expanded, and Engagement indicator engagement gradually rose. In three months, engagement work expanded from three types to seven, and AI-processed call volume expanded to 2.4 times the initial level.
As a secondary effect, night shift restructuring became possible. Late-night phone duty became unnecessary, allowing reduction of one night shift staff. Personnel costs from the reduction were redirected to daytime front desk reinforcement, raising service quality during busy hours. "The shift overall changed to a design concentrated on the dense customer service hours," the report noted.
The end of Iwase's report read: "Even members weak with IT take root if usage scenarios are limited. Following stages with HEART, it entered the field without resistance. Tools that take root are designed to take root."
It was the day the front desk could look only at the guests in front of them.
"Tool implementation succeeds not at the moment of launch, but at the moment of adoption. What HEART asks is the five indicators of satisfaction, engagement, adoption, retention, and success. Even with high adoption rate, low satisfaction causes drop-off. Even with high task success rate, without expanding engagement breadth, investment isn't activated. By tracking five indicators simultaneously, the contour through adoption becomes visible. The day the switchboard releasing the front desk from guests disappeared, the tool was quietly settling into the field."
Related Files
Tools Used
- ROI Polygraph — Visualizing phone duty workload, interruption cost, and missed calls
- ROI Proposal Generator — Investment recovery simulation for staged AI operator implementation
