ROI Case File No.591: They Heard AI Could Do Anything, but Had No Yardstick for What to Aim at First
They Heard AI Could Do Anything, but Had No Yardstick for What to Aim at First
Chapter 1: They Want AI to Change Sales, but Can't Settle on Where to Aim
"I want to use AI to change the way we sell, from the ground up."
Isao Warida, sales division manager at Titan Solutions, said this as he laid out the situation. "We use Google Workspace company-wide. I want to layer Gemini Enterprise and Google Antigravity on top of it and get AI into the hands of all fifty of our salespeople. At trade shows and in case studies, I've heard about mountains of things it can do."
"Among all of that, what do you want to do?" Claude asked.
"All of it," Warida answered. "Automatic meeting notes, first drafts of proposals, organizing customer information, drafting emails, lost-deal analysis. I have the things I want to do listed on paper. But the company is split on which to aim at first. The executives say close rate, the floor says the burden of note-taking, IT says the safety of the integrations. Every one of them argues something different."
"Have you laid those initiatives side by side and compared them by reach, impact, and effort?" I confirmed.
"...We haven't compared them," Warida answered. "All we've done is list the things we want to do. How many people each one reaches, how much it moves per person, how much effort it takes. We've never put them on the same yardstick."
"A list alone doesn't produce an order," I replied. "Let's break this down with RICE."
Chapter 2: RICE Asks—Divide the Impact by the Effort
"This case calls for RICE."
Claude wrote "Reach / Impact / Confidence / Effort" on the whiteboard.
"RICE—Reach, Impact, Confidence, Effort—is a framework that scores initiatives on four yardsticks, then multiplies and divides to line them up in a single row," I explained. "The crux is not letting the argument be settled by whoever pushes hardest. How many people does it reach, how much does it move per case, how solid is that read, and how much effort does it take? Multiply the first three and divide by the last, and the initiatives line up by score rather than by preference."
"First, let's measure the current cost," Gemini said, opening ROI Polygraph. The data Warida had provided went in.
"The monthly cost is out," Gemini read off. "Labor for recording meeting notes, preparing proposals, and organizing customer information averages 175 hours a month; at ¥3,900 an hour, that's ¥682,000 a month. Opportunity loss from sales know-how staying locked inside individuals averages ¥400,000 a month. Delayed deployment decisions caused by never settling the priority of AI initiatives average ¥360,000 a month. The inability to analyze lost-deal causes, because meeting records aren't maintained, averages ¥320,000 a month. Waste from duplicate tool contracts and scattered trials averages ¥300,000 a month. Total: ¥2,062,000 a month. Annualized, roughly ¥24,740,000."
Warida stared at the figures. "I was only looking at the data-entry burden. Once you add the delay from decisions that never get made, and the cost of tools we tried and abandoned, it comes to this much."
"Then let's design it with RICE," I continued.
[Reach—count how many people it touches, and how often]
"First, reach," Claude said. "Automatic meeting notes touch all fifty salespeople every day. It rides on more than a thousand meetings a month. Lost-deal analysis, on the other hand, runs a few times a month and is used by only a handful of planning staff. Even for the same kind of function, when reach differs by two orders of magnitude, the order changes."
[Impact—see how much changes per case]
"Next, impact," Gemini continued. "How many minutes are saved per case, how many more calls can be made. Automating notes saves twenty minutes per meeting. A proposal draft saves an hour per case. The size of the impact only means something once it's multiplied by reach."
[Confidence—discount the read by how solid it is]
"After impact comes confidence," I continued. "Since Google Workspace is already in place, the read on integration and migration is solid. But the read that AI will lift close rate by twenty percent has no grounding. Discount by confidence, and the solid initiatives rise while the ones built on hope alone fall."
[Effort—divide the product by the effort]
"Last, effort," Claude continued. "However well something works, if it takes six months the score drops. Because you divide the product by effort, the things that can ship small and early naturally come to the front. Don't design a company-wide sales transformation all at once—ship in order of the divided score."
[Simulating the investment recovery]
"Let's run the numbers with ROI Proposal Generator," Gemini proposed.
- Initial cost: Initiative inventory and RICE scoring, Gemini Enterprise deployment design, integration with existing Workspace, implementation of meeting notes and proposal generation, and sales training—¥4,800,000 in total
- Monthly cost: Licenses plus ongoing operation and maintenance—¥230,000 a month
- Monthly savings: Automation of meeting notes and minutes = ¥480,000 a month (assuming a 70% reduction), reduced preparation labor through generated proposal drafts = ¥340,000 a month, suppression of opportunity loss through know-how sharing = ¥280,000 a month, cleanup of duplicate and trial tools = ¥240,000 a month—¥1,340,000 a month in total
- Net monthly savings: ¥1,340,000 − ¥230,000 = ¥1,110,000 a month
- Payback period: ¥4,800,000 ÷ ¥1,110,000 = approximately 4.3 months
"That's a payback of a little over four months," Gemini summarized. "What works is that you stop listing the things you want to do and start lining them up by a score you multiply and divide. Start with the case made by whoever argues loudest, and you spend effort on initiatives that reach almost no one. Line them up by score, and the thing that touches all fifty people every day comes first. The investment doesn't swing at air."
Warida looked over the numbers. "I thought it was enough to pick a good AI. Without scoring, you can't even decide what to aim at first."
"RICE is a tool for dividing impact by effort and lining things up in a single row," I replied.
Chapter 3: A Deployment Plan That Ships in Order of Score
"Let me lay out the approach," I said, standing at the whiteboard.
"Month one—inventory the initiatives and score them on the four yardsticks. Month two—fix the scores and decide the implementation scope and target departments. Months three and four—implement automatic meeting notes and the Google Workspace integration. Month five—roll out proposal-draft generation and train the sales force. Month six—integrate customer-information organization and verify results. Month seven onward—revise the scoring against measured values, and begin lost-deal analysis."
"Does that mean the low-scoring initiatives simply never happen?" Warida confirmed.
"Not never—later," Claude replied. "Lost-deal analysis has too little material to work with if meeting notes haven't accumulated; done now, it won't be accurate. Run the high-scoring note automation first and the material for the analysis piles up on its own, so that in six months the effort drops and the score rises. Deciding an order is not the same as discarding."
Warida took notes as he spoke. "Decide the order of attack before choosing the AI. I can see the sequence now."
Chapter 4: The Day Scores Lined Up and the Company Agreed
Ten months later, a report arrived from Warida.
Meeting notes changed immediately after implementation. "What used to take an hour of writing from memory after getting back from a visit now takes only a check and a few additions. Every day got lighter for all fifty of them," Warida wrote.
Proposals got faster too. Drafts built on past deals eliminated the blank page. "The time spent writing from zero disappeared. Salespeople started thinking about substance instead of structure," the report said.
The largest change showed up in how the company made decisions. A state of fighting over who argued hardest became a state of lining things up by score. "The executives, the floor, and IT all said different things. Once we scored on the four yardsticks, the discussion ended in thirty minutes. Even the side that had opposed it accepted the scores," Warida wrote.
Lost-deal analysis started moving as well, once its time came. The accumulated meeting notes became the material directly. "If we'd done it earlier, we'd have spun our wheels with nothing to analyze. Making it wait its turn paid off," the report said.
As a side effect, tool contracts got cleaned up. The habit of trying things and abandoning them stopped. "We stopped bringing things in because they looked interesting. We started asking how many people it reaches and how much it moves," Warida wrote.
The final line of Warida's report read: "I thought the trouble with AI was that we couldn't find good tools. But the real problem was that we'd heard AI could do anything, and had no yardstick for what to aim at first. The moment we scored the four with RICE, both the front of the line and the back were settled. Before choosing tools, deciding the order came first."
The day a company with no yardstick for what to aim at first became a company that could decide by score, AI deployment in sales had changed from listing the things they wanted to do into a design that divides reach, impact, and confidence by effort to line everything up in a single row, the report noted.
"Consultations about AI usually arrive in the form of 'I want to know what it can do.' But there is a question to ask before choosing a tool. Has it been decided what to aim at first? What RICE asks is reach, impact, and confidence—and the effort that divides them. Multiply and divide, and initiatives line up by score rather than by the force of an argument. The day a company that had been listing wishes could decide an order, what changed was not the performance of the AI but the very perspective of dividing impact by effort and comparing."
Related Files
Tools Used
- ROI Polygraph — Visualizing labor for meeting notes and proposals, opportunity loss from personalized know-how, and decision delay from unsettled priorities
- ROI Proposal Generator — Investment-recovery simulation for sales AI deployment starting from scoring on four yardsticks
