ROI Case File No.453 'Seventy Frames, Repeated'
Seventy Frames, Repeated
Chapter 1: The Weight of "A Little"
"It takes a little time, and I was wondering if there's anything we can do."
Kenichi Kirishima, manager at AutoVantage, said this while scratching his head. At this used car dealership on the outskirts of Hokkaido, three staff members were taking turns sitting in front of screens, repeating the same task.
"It's the vehicle registration process," Kirishima explained. "Every time we take in a car, we photograph it and upload the images to the listing site. Right now, we drag and drop each image one by one into a set form. Around seventy images per vehicle, all done by hand."
"How long does it take per vehicle?" I asked.
"About fifteen, twenty minutes," Kirishima answered.
Claude asked quietly: "How many vehicles do you register per month?"
"Lately, up to about fifty a month."
"Fifty vehicles, averaging seventeen-and-a-half minutes each—" Gemini calculated. "That's eight hundred and seventy-five minutes a month. Roughly fourteen-and-a-half hours."
Kirishima's eyes went wide. "That much?"
"That's the number the word 'a little' was hiding," I said. "Fifteen hours a month, staff tied up on registration. Customer-facing time stopped while that's happening. That number is the real question behind today's request."
Chapter 2: Turning Vague Words into Numbers
"This case calls for BOM."
Claude wrote on the whiteboard: BOM — Basis of Measurement.
"BOM is a framework for scaling ambiguous language and converting it into concrete numbers," I explained. "The phrase 'it takes a little time' means something completely different to different people. The manager's 'a little' and a staff member's 'a little' are probably not the same. Without accurate numbers, there's no ROI calculation, and no grounded improvement proposal."
"Let's measure the current state accurately first," Gemini proposed. "This week, time every registration job with a stopwatch, vehicle by vehicle. Record the variation for cases with more or fewer than seventy images."
A week later, Kirishima brought the data. It was entered into ROI Polygraph.
"Here are the results," Gemini said, reading from the screen. "Registration time per vehicle: minimum twelve minutes, maximum twenty-six minutes. Average: 18.3 minutes. Image count ranged from sixty-two to eighty-one. The primary driver of time variation: file naming. Because file names from the shoot are inconsistent, staff need to verify which photo goes in which field every single time."
"In other words," I summarized, "the bottleneck isn't the drag-and-drop of seventy images. It's the file name verification and sorting. The location of the problem has shifted."
Kirishima looked up. "Now that you say it — the shoot order and the form input order don't match. You have to check as you go or you'll make mistakes."
[Estimating RPA Impact with Numbers]
"You mentioned considering RPA," Claude continued. "With BOM numbers in hand, we can estimate the impact accurately. RPA handles rule-based repetitive tasks. This problem has two layers."
"Layer one — standardizing the photography stage. Establishing a rule to automatically assign vehicle number and shot type to each file name eliminates the verification step downstream. This is a change in operating procedure, not a technology implementation. It can be done immediately."
"Layer two — RPA for the registration task itself. With standardized file names, RPA can complete registration per vehicle in under three minutes. From eighteen minutes down to three — an eighty-three percent reduction."
"At fifty vehicles a month," Gemini added, "monthly registration time drops from nine hundred and fifteen minutes to one hundred and fifty minutes. The difference is seven hundred and sixty-five minutes — about twelve-and-a-half hours. That time comes back to customer-facing work."
Kirishima wrote down the numbers, then said quietly: "Thirteen hours coming back every month. That's enough for one staff member to add roughly half a month's worth of customer contact time."
Chapter 3: Numbers Move Decisions
"What's needed for the RPA decision," I continued, "is the cost-versus-impact comparison. Let's model it in ROI Proposal Generator."
Workforce savings in labor cost terms, RPA implementation cost, running cost — the estimate returned. Payback period: seven months. With one condition attached: Layer One, the photography standardization, must be completed first.
"This is the critical point," Claude emphasized. "Before bringing in RPA, unify the file naming convention at the shoot. Without it, the data RPA reads is different every time — errors pile up. Two weeks for the Layer One workflow change, then into RPA implementation. Observe that sequence and the seven-month payback holds."
"Layer One alone produces results," Gemini added. "Simply organizing the file names eliminates the verification step, and registration time drops from eighteen minutes to around twelve. This is an improvement that can start today, without waiting for RPA."
Kirishima folded his arms in thought. "Right — the file naming is something we can change today. I'll talk to the staff and start from the shooting rules."
"That's the power of BOM," I said. "Converting 'it takes a little time' into a number shows you what to tackle first. Act on instinct and you want to start with the big RPA implementation. Look at the numbers and you see the first move is changing the shoot naming rule."
Chapter 4: Beyond the Seventy Frames
As Kirishima stood, he said offhandedly:
"I haven't thought about what the staff will do when registration time decreases. When the time comes back — if we haven't decided what to use it for, it'll just get swallowed up by something else."
"Exactly," I replied. "Efficiency is not the goal. The real question is what you create with the time you recover. If thirteen hours returns to customer-facing work per month, how many additional appointments does that generate? How many additional sales? BOM is the foundation that makes that calculation possible."
Outside the window, used cars lined up in the parking lot caught the winter light.
Four months later, a report arrived from Kirishima.
The photography rules were changed in week one. File naming standardization alone reduced average registration time to 11.2 minutes. Two months later, RPA was introduced and registration time compressed to 2.8 minutes.
Monthly registration workload dropped from nine hundred and fifteen minutes to one hundred and forty. The recovered time was redirected to follow-up calls with existing customers and test drive invitations. Over four months, the closing rate improved eleven percent compared to pre-implementation.
Kirishima's report read: "Until I converted 'it takes a little time' into a number, I hadn't realized what a significant problem it was. Converting it to a number was the first step to seeing what the problem was and what to tackle first. BOM is a framework for translating language."
Beyond the seventy frames, different work had been waiting all along.
"Operational problems hide inside vague language. 'It takes a little time' is an impression. Until you measure it, you can't know the size of the problem, the order of intervention, or the value of the investment. What BOM asks is: how many minutes? how many cases? how many yen? The moment you convert it to a number, the problem stops wearing the face of a problem and starts wearing the face of a solvable challenge. The habit of scaling ambiguity generates the right first move."
Related Files
Tools Used
- ROI Polygraph — Measurement data analysis and identification of workflow bottlenecks
- ROI Proposal Generator — RPA implementation payback simulation
