ROI Case File No.403: The One Who Traces the Value Chain
The One Who Traces the Value Chain
Chapter 1: The Truth of Distorted Matching
"Do job sites truly connect job seekers with companies?"
The founder of MatchMakers opened with this question. In her hand, she gripped printed materials showing search results from major job sites.
"Look at this. The top-ranked results are all companies that paid more for advertising. Job seeker skills and aptitude, company realities and culture—elements that should be core to matching have absolutely no bearing on display order."
Her voice carried clear awareness of the problem.
"I previously worked at a staffing agency. There I repeatedly witnessed job seekers disappointed by gaps between glamorous job ads and reality. And the vicious cycle of companies incurring recruitment costs again when hired personnel quickly resign."
"What is true matching? I want to create a new recruitment service pursuing that."
Her vision was a completely new recruitment platform using AI. Analyzing job seeker aptitude and company realities, matching by compatibility rather than advertising fees. AI analyzing employee and former employee reviews to visualize company realities. And transforming job hunting itself with intuitive UI like dating apps.
"However," she hesitated, "where to start, how much functionality to implement initially. The implementation method for AI-based review analysis also isn't concretely clear."
A grand vision with an unclear execution plan—that was her contradiction.
Chapter 2: Where Value Is Born
"For this case, the VALUE CHAIN Model is suitable."
Gemini drew a horizontal arrow on the whiteboard with several boxes lined up above it.
"The VALUE CHAIN Model," I began explaining, "is a method that breaks down the series of activities through which a company provides value to customers, visualizing where value is created and where it's lost."
"MatchMakers' vision is grand, but let's organize that overall picture as a 'flow of value.'"
Claude asked the founder, "What do you think is the 'value' job seekers ultimately receive?"
The founder answered immediately, "Meeting companies suited to them. Workplaces where skills are utilized and culture fits. The fulfillment and growth gained from working there."
"Then through what chain of activities is that value created?"
I began writing the value flow on the whiteboard.
"First, job seeker information collection. This is the starting point."
[Step 1: Job Seeker Data Collection and Analysis]
"Job seeker skills, experience, values, work style preferences—without accurately grasping these, matching cannot begin," Gemini organized.
"Current job sites judge this solely from resumes and work histories," the founder added. "But true aptitude must lie much deeper."
"That's where AI comes in," Claude continued. "Through questionnaire-style aptitude assessments and analysis of past behavior patterns, visualizing aptitudes job seekers themselves haven't noticed."
[Step 2: Company Information Collection and Reality Analysis]
The next arrow pointed toward companies.
"We need to gather not just official company information, but real voices from employees and former employees," I pointed out. "Here's where AI analysis of review information becomes crucial."
The founder leaned forward. "Exactly. But how to implement it?"
"Using natural language processing, specifically sentiment analysis and topic extraction techniques," Gemini began explaining. "From review texts, automatically extract themes like 'overtime frequency,' 'relationship quality,' 'growth opportunities,' and determine whether each is positive or negative."
"However," I supplemented, "no need to aim for perfection initially. First focus on limited topics, increasing accuracy. Gradually expand scope from there."
[Step 3: Matching Algorithm Construction]
The value chain advanced to the next step.
"Once job seeker aptitude data and company reality data are ready, an algorithm connecting them is needed," Claude organized.
"Not simple keyword matching, but multidimensional compatibility determination," the founder confirmed.
"Exactly. And what's crucial," I emphasized, "is building from the start a mechanism to learn this algorithm's accuracy from actual matching results. Meaning, accumulating data on 'successful matches' and 'failed matches' to continuously improve the model."
[Step 4: User Experience Design]
The final stage of the value chain was actual user experience.
"However superior the matching algorithm, value doesn't transmit through difficult UI," Claude pointed out.
"Intuitive operation like dating apps," the founder said. "The image of swiping companies left or right to select 'interested' or 'not interested.'"
"That's an excellent idea, but," Gemini cautioned, "in early stages, even simple list display and detail screens suffice. Swipe UI can wait until basic functionality reproducibility is confirmed."
Chapter 3: Priority Within the Chain
"Drawing the VALUE CHAIN Model clarifies the overall picture," the founder gazed at materials. "But doubt remains. Must all four steps be perfectly implemented simultaneously?"
I quietly shook my head. "No. The VALUE CHAIN Model's true value lies in discerning 'where the most important value creation point exists.'"
"In MatchMakers' case, which step do you think provides the most differentiation?"
After brief thought, the founder answered, "Company reality analysis. The part that AI-analyzes reviews to visualize company culture and work environment."
"Why do you think so?"
"Because that's what existing job sites most lack. Other services do job seeker aptitude assessment, but few objectively show company realities."
"Exactly," I nodded. "Then initially, focus there."
Gemini organized. "Meaning, Step 1 job seeker analysis starts from simple questionnaire level. Step 3 matching algorithms also start from basic keyword matching and recommendations. But concentrate resources on Step 2 company reality analysis."
"And Step 4 UI is also sufficient with simple search and list display," Claude continued. "Elaborate features like swipe UI come after core functionality reproducibility is proven."
The founder's expression brightened. "No need to perfect everything from the start."
"Exactly," I answered. "The VALUE CHAIN Model visualizes the value chain. But what's important is discerning where the 'bottleneck' and 'differentiation point' exist within that chain."
Chapter 4: From Small Chains to Great Flows
"Then what's the concrete first step?" the founder asked.
I proposed, "First, start service experimentally focused on specific industries or job types. For example, specialize only in IT industry engineer positions."
"There are two reasons to narrow scope. One is easier optimization of review analysis AI models to industry-specific terminology and context. The other is easier success case creation."
Gemini supplemented, "Achieving high-accuracy matching in a narrow scope becomes a model case when expanding to other industries."
"And what's crucial," Claude added, "is thoroughly tracking and recording the first several dozen matching results. Whether job seekers actually joined, whether they're satisfied after joining, whether companies are also satisfied with hiring."
"That data becomes the foundation for algorithm improvement," the founder showed understanding.
"Exactly. Each VALUE CHAIN step isn't independent but connected through feedback loops. Actual matching results connect to improving job seeker analysis accuracy, improving company analysis, and optimizing algorithms."
The founder stood and bowed deeply. "The destination is clear. And the first path there as well."
After she left, Gemini said, "The VALUE CHAIN Model is a method to understand the whole by breaking down activities."
"Yes," I answered. "But breaking down isn't the purpose. The essence is understanding the 'connections' of each broken-down step and discerning where to apply effort."
"And," Claude continued, "in that place requiring effort, first prove reproducibility small."
Outside the window, twilight light illuminated the office.
Six months later, a report arrived from MatchMakers. In the trial version service specialized for IT industry, they implemented the first 50 matches. Of these, 42 actually resulted in joining, and in three-month satisfaction surveys, 85% of job seekers and 80% of companies responded "satisfied."
Review AI analysis accuracy had also greatly improved from the initial version.
The value chain had begun small and was flowing surely.
"Value is born within chains. Discerning where in that chain to focus effort and proving reproducibility there small—that's the only path to transform grand visions into reality."
