ROI Case File No.583: They Had Decided How Much to Sell, but Not Who They Were Selling To
They Had Decided How Much to Sell, but Not Who They Were Selling To
Chapter 1: They Want to Reach Ten Million, but the Moves Won't Settle
"We want to lift our EC monthly revenue to ¥10,000,000 within six months."
Akira Hitomi, CEO of QuantumNeuro, said this as he laid out the situation. "Right now we bounce between five and seven million a month. It ticked up after we replaced the 3D display and AR features. To jump another level from here, I think the only way is combining EC with big data and AI."
"What kind of combination are you considering?" Claude asked.
"A CRM that links an AI agent with BigQuery," Hitomi answered. "And on the entry side, I want AI agents and chatbots for customer acquisition. Honestly, though, we still don't have an efficient pattern for acquiring customers. We have the data, but we can't turn it into strategy."
"Have you decided who buys, such that the ten million accumulates?" I confirmed.
"...We haven't," Hitomi answered. "All I've thought about is the revenue target. What kind of person, in what situation, buying for what reason. I've never narrowed the other party down to a single person."
"Deciding how much to sell doesn't let you choose your moves if you haven't decided who you're selling to," I replied. "Let's break this down with persona analysis."
Chapter 2: Persona Analysis Asks—Narrow the Customer Image to One Person
"This case calls for persona analysis."
Claude wrote "Attributes / Context / One Person" on the whiteboard.
"Persona analysis is a framework for drawing the buyer as one concrete individual and building your tactics toward that person," I explained. "The essence is not trying to sell to everyone. An ad aimed at everyone pierces no one. Layer the reason for buying onto attributes like age band and purchase history, and narrow it down to a single face. Once the other party is decided, the ads, the content, and the AI's replies all follow naturally."
"First, let's measure the current cost," Gemini said, opening ROI Polygraph. The data Hitomi provided went in.
"The monthly costs are out," Gemini read off. "Wasted ad spend from running campaigns without a settled customer image averages ¥480,000 per month. Opportunity loss from depressed CVR due to uniform delivery to all customers averages ¥440,000 per month. Purchase-data analysis and report preparation averages 140 hours per month; at ¥3,600 per hour, that's ¥504,000 per month. Labor from handling inquiries manually averages ¥320,000 per month. Churn from the absence of repeat-purchase tactics averages ¥340,000 per month. The total is ¥2,084,000 per month. Annualized, roughly ¥25,000,000."
Hitomi stared at the figures. "I was only looking at revenue not growing. Once you add the waste from advertising scattered without deciding who we're selling to, it comes to this much."
"Then let's design with persona analysis," I continued.
[Attributes—Draw the outline of the people already buying]
"First, attributes," Claude said. "Purchase history, age band, acquisition channel, products viewed. Draw the outline from the data on people who have already bought. Don't invent a persona from imagination. The data in your hands is the first line."
[Context—Write in why they buy and in what situation]
"Next, context," Gemini continued. "People who buy after viewing 3D and AR—what are they trying to confirm? Where it will fit, the texture, how it will look at the recipient's home? Write in the reason and the situation. Attributes alone don't decide the words you use."
[One Person—Discard the others and narrow to one]
"After context, the narrowing," I continued. "Even if three or four candidates emerge, the first target is one. Choose by contribution to revenue and ease of acquisition. It's painful to discard the rest, I know, but without narrowing, your ad copy and your chatbot's replies can only ever be inoffensive."
[Implementation—Load the persona into the CRM and the AI agent]
"Last, implementation," Claude continued. "Drop the one person you narrowed to into BigQuery extraction conditions and make it a CRM segment. Design the AI agent's and chatbot's responses toward that person too. The persona becomes the configuration value of the mechanism itself."
[Calculating the payback]
"Let's run the numbers with ROI Proposal Generator," Gemini proposed.
- Initial cost: Purchase-data analysis, persona design, CRM build and BigQuery integration, AI agent/chatbot implementation, and content production — ¥5,000,000 total
- Monthly cost: Ongoing operation of the CRM and AI platform — ¥230,000 per month
- Monthly savings: Reduced ad waste through targeted placement = ¥500,000 per month (assuming a 70% reduction), CVR improvement through personalization = ¥360,000 per month, automation of analysis and reporting = ¥300,000 per month, reduced inquiry handling via chatbot = ¥260,000 per month — ¥1,420,000 per month in total
- Net monthly savings: ¥1,420,000 − ¥230,000 = ¥1,190,000 per month
- Payback period: ¥5,000,000 ÷ ¥1,190,000 = approximately 4.2 months
"A payback of just over four months," Gemini summarized. "What makes it work is not running on a revenue target alone, but narrowing the other party down to one person. Without a decided target, ads scatter in all directions and the AI agent can only return generalities. Because the persona is decided, the same budget lands somewhere different. The investment doesn't swing at air."
Hitomi looked over the figures. "I thought putting in good machinery would grow it. Without deciding who we're selling to, the machinery misses the mark too."
"Persona analysis is a tool for narrowing the customer image to one person," I replied.
Chapter 3: An Implementation Plan That Narrows to One
"Let me lay out how to proceed," I said, standing at the whiteboard.
"Month one—analyze purchase data and generate candidate customer images. Month two—interview for context and fix the one person to target. Months three and four—build the CRM, integrate BigQuery, and implement the segments. Month five—design AI agent and chatbot responses and produce content. Month six—run the campaigns and verify results. Month seven onward—expand to a second persona and keep the customer image updated."
"Doesn't narrowing mean giving up revenue from everyone else?" Hitomi asked.
"That's the point most often misunderstood," Claude replied. "What you narrow is the aim, not the market. Words aimed at one person have a sharp outline, so they reach the people around that person too. Conversely, words aimed at everyone pierce no one and only drain the ad budget. Narrowing and then widening comes before widening from the start."
Hitomi took notes. "Before deciding the amount, decide the person. I can see the order now."
Chapter 4: The Day the Other Party Was Decided and the Numbers Moved
Ten months later, a report arrived from Hitomi.
Advertising changed dramatically once the persona was fixed. "Same budget, different landing point. Acquisitions per yen went up compared with when we were scattering," Hitomi wrote.
The AI agent's responses changed too. Designing for the one target sharpened the conversations. "We stopped giving safe answers that work for anyone, and pre-purchase questions started converting straight into orders," the report said.
The largest change showed up in how tactics were decided. From holding up a revenue target alone, they moved to deciding the other party and building from there. "I was just saying ten million. Once I wrote down who buys such that it becomes ten million, the work lined itself up," Hitomi wrote.
Repeat purchases started moving as well. Designing return visits along the context stopped the churn. "We knew what they'd want to know after buying, so what to deliver was decided. Second purchases increased," the report said.
As a secondary effect, their view of data changed. The habit of translating data into a persona, rather than merely collecting it, took root. "We used to stop at 'we have big data.' Now we ask what this data shows about whom," Hitomi wrote.
At the end of Hitomi's report, he had written: "I thought the EC problem was that revenue was short. But the real problem was that we had decided how much to sell and not who we were selling to. The moment persona analysis narrowed it to one person, the ads, the content, and the AI's replies were all decided. Before deciding the amount, deciding the person came first."
The day a company that had decided how much to sell without deciding who became a company that narrows to one and builds accordingly, EC revenue growth had changed from setting a target figure into a design that narrows the customer image to one person before building the tactics, the report noted.
"Revenue-growth consultations almost always arrive in the form of 'we want monthly revenue at this figure.' But before you hold up the number, there is something to ask. Who buys it? What persona analysis asks about is attributes, context, and the one person you narrowed to. Once the other party is decided, the ads, the content, and the AI's words follow on their own. The day a company that had decided only how much to sell managed to decide who to sell to, what changed was not the height of the target but the very perspective of narrowing the customer image to one person."
Related Files
Tools Used
- ROI Polygraph — Visualizing ad waste from an absent customer image, depressed CVR from uniform delivery, and analysis labor
- ROI Proposal Generator — Investment-recovery simulation for AI-driven EC revenue growth tactics starting from narrowing the customer image
