Professional service firms, such as those for tax accounting or labor and social security, are businesses where management decisions tend to rely heavily on “the president’s intuition” or “years of experience.” To resolve this “personalization,” services known as “The President’s Digital Twin” have emerged, aiming to visualize and structure decision-making. At first glance, this appears to be the practice of “reversible management”—systematizing personal knowledge to create a reproducible decision-making process. It seems like an ideal solution.
However, there is a major pitfall here. It is the risk that simply replacing “personalized decisions” with “decisions fixed within a system” can actually lead to a loss of reversibility and create new rigidity. Tool implementation must always be designed as a “reversible experiment.” Using this service as a case study, we will consider how to ensure a “reversible design” when structuring decision-making.
Are You “Fixing” Decisions Under the Guise of “Organizing”?
The goal of services like “The President’s Digital Twin” is to convert tacit knowledge into explicit knowledge. This is the right direction. The problem lies in the “method.” The trap many tool implementation projects fall into is the approach of “digitally transferring personalized tasks exactly as they are.” The decision criteria in the president’s mind are extracted through interviews and hearings, then embedded into the system as rules and flows.
At this point, the manager must ask: “Is this decision process I am currently using truly optimal? Or is it merely a ‘local rule’ born from past habits and constraints?” Simply fixing personalized decisions as they are is a dangerous act that “legitimizes” and institutionalizes inefficiency and irrationality. It may be equivalent to turning the “soft chains” of personalization into the “hard cage” of systematization.
Three Points Where Reversibility is Lost: The Trap of Tool Dependency
When delegating decisions to a system, there are three main points where reversibility can be lost.
1. The “Black-Boxing” of Decision Logic
The decision algorithms or recommended flows provided by such services are often based on external know-how or general best practices. Applying these before deeply observing and verifying your own company’s reality leads to a shallow understanding of “why that decision was reached,” fostering an organization that blindly follows the tool’s output. The tool becomes the entity that provides the “correct answer,” and members degrade their ability to think for themselves and adjust judgments according to the situation—the very essence of a professional service firm’s value.
2. The “Tool-Dependent Hardening” of Workflows
Once a workflow is optimized for a tool, it becomes difficult to change. The tool’s screen transitions and input fields dictate the work flow, leading to a state of mental stagnation where “what the tool cannot do, we won’t do in our work either.” There is a particular danger that the nuanced differences between clients or handling exceptional cases—areas where the true value of a professional service is tested—get discarded. The tool risks “simplifying” operations rather than “making them efficient.”
3. Loss of Learning and Adaptation Opportunities
Even if personalized decision processes contain waste or inefficiency, they hold the historical context and traces of trial and error that explain “why they became that way.” If this is “organized” away all at once, there is a risk of losing the “organizational memory” needed for the company to learn from past failures and adapt to changing circumstances. When all decisions are delegated to a tool, the muscle for thinking independently and experimenting atrophies when facing new challenges.
Four Designs for a “Reversible” Tool Implementation
So, how can we advance the structuring of decisions while eliminating personalization and preserving reversibility? Clarify the following four points during the experimental design phase.
Design 1: Establish a Period for “Partial Implementation” and “Parallel Operation”
Do not migrate all operations at once. Select one operational area that is highly personalized yet relatively low in importance (where the cost of failure is small), and run tool-based decisions and traditional human decisions in parallel. This “parallel operation period” is not just a testing phase but a valuable “observation period.” Where do the tool’s outputs and human judgments align, and where do they diverge? Is the divergence due to a tool shortcoming or human bias? Maintain this state for at least three months, ideally six, to accumulate data and insights.
Design 2: Treat it as a “Provisional Guideline,” Not an “Absolute Rule”
Position the decision flows and criteria suggested by the tool not as “absolute rules to be obeyed,” but as “provisional guidelines to follow initially.” Then, clearly empower members: “If you feel a judgment deviating from the guideline is necessary, you must record the reason, the action taken, and the result, and review it weekly with the team.” This creates a cycle for discovering the tool’s limits and complementing/updating it with organizational knowledge.
Design 3: Define Exit Conditions Based on “Quality of Judgment,” Not “Usage Rate”
Do not mistake the metrics for measuring the success of tool implementation. “100% tool usage rate” is not success; it might even be a danger signal of blind obedience. What should be evaluated are substantive outcomes like: changes in “decision speed and quality,” “whether members can explain the rationale for decisions,” and “whether responsiveness to exceptional cases has declined.” For exit conditions, incorporate qualitative observations, such as: “After the parallel operation period ends, if two or more out of three key members feel ‘the decision burden has decreased, but flexible response has become harder,’ we will review the implementation scope.”
Design 4: Build an “Override Function” into the Tool
If technically possible, make an “override with reason” function for the tool’s judgments mandatory. This override data is the most valuable material for customizing the tool to your company’s reality. By accumulating and analyzing data on why an override occurred and what the result was, you can highlight the “truly valuable tacit knowledge” that was personalized and leverage it for the next system improvement. The tool should not be a provider of “answers” but a “dialogue partner for better judgment.”
Conclusion: The “Twin” Should Be a “Questioner,” Not “Another Self”
The naming “The President’s Digital Twin” is intriguing but contains danger. The twin must not be “another rigid version of yourself.” The ideal twin should be “an entity that constantly asks ‘Why?’, ‘Really?’, ‘What else?’ and prompts you to think more deeply and broadly about your decisions.”
Resolving personalization must not be an “abandonment” or “outsourcing” of judgment. It should be about the “structuring” of judgment and the “promotion of reflection.” Tool implementation is not about making decisions easier; it is an investment in making them better.
When considering such services, professional service firm leaders must ask: Will it become a “coffin” that fixes their judgment, or a “mirror” that hones their decision-making skills? The turning point lies in whether the implementation can be designed as an “observable experiment,” not a “finished product.” Rather than seeking a perfect system from the start, you should first take a step forward and, before even choosing a tool, clearly map out a reliable path to return if that step proves to be a mistake.
Comments