Customers do not see your org chart.
Imagine an appliance subscription brand promising next-day installation in an ad. Checkout shows three to five days in some regions, while the support bot repeats the ad. This is an illustrative scenario, not a documented company case. Polishing the copy will not create installation capacity. The company needs to decide what it can promise, where, using the same current facts across the business.
Part I described Brand Design Management as a system that connects organizational choices around brand strategy. Chapter 9 of my Korean book proposes BDM-OS to make that work repeatable. Its four layers are not four new departments. They are four kinds of work through which a promise is made, delivered, checked and carried into the next decision. (Korean print edition, pp. 92–97, 104)
Test one promise across four layers.
| BDM-OS layer | Work in the installation scenario | What to retain |
|---|---|---|
| Brand strategy | Replace “next day everywhere” with a promise to show the actual installation date before the customer commits. | The customer, the scope of the promise and conditions for changing it |
| Organizational integration | Operations checks capacity by region; product connects it to checkout; marketing and support use the same current source. | Decision owners, exceptions and a route for challenge |
| Artwork governance | Check that visual emphasis in ads and checkout does not bury timing exceptions. | Expression criteria, approvals, originals and usage rights |
| Brand knowledge management | Record why the conflicting promises arose and whether the correction worked. | Sources, versions, reasons for decisions, customer response and open questions |
This table applies the four layers described on pages 94–97 and 104 of the book to a hypothetical situation. It is neither a case from the book nor a validated recipe. The point is to connect each expression to the operating facts that can support it.
Who decides, and who can challenge?
Operations verifies when installation is possible. Product and service design what customers see before payment. Marketing and artwork specialists shape the expression. The company must assign authority for the promise itself: who approves “next day” for a given region, and who can withdraw it when capacity changes? Listing teams in a table does not assign a final decision owner. (pp. 95, 100–101)
A Brand Design Management leader need not approve every image or replace technical judgment. Specialists should be able to present evidence and challenge a proposal within their fields. The leader makes conflicts visible and records the criteria and authority behind the resolution. In a small company, existing leaders may share this role; a larger company may need decision forums at different stages. The book does not prescribe one org chart. (pp. 100–101)
What review does an AI-generated answer need?
AI can group regional questions, flag contradictions between old ads and current installation rules, and draft alternative wording. Yet a summary based on last month’s availability differs from one that handles private customer data. The book proposes different review levels for public-information drafts, strategic or creative options, and decisions involving price, quality, legal duties or confidential information. The work’s risk, rather than a fixed number of approvals, should determine the check. (pp. 97–99)
Before a support bot publishes a new answer, someone must check the latest authoritative availability data and the regions it covers. Higher-risk changes require the relevant specialists and an authorized decision maker. Even a cited AI answer needs its source checked for existence, currency and fit. The proposal is to define trustworthy inputs and verification responsibilities for the work AI assists. (pp. 98–100)
Start with one project, then review it.
The book recommends starting with one brand and one recent project rather than installing a company-wide system at once. In this illustrative scenario, a team might proceed as follows. (p. 102)
- Collect the memory: gather the ad, regional capacity records, checkout, support answers and prior decisions with their dates and sources.
- Align the criteria: define what can be promised to whom and what operations, product, marketing and artwork must each verify.
- Try one project: test decision rights, exception notices, data updates and AI-assisted work for one region or product.
- Review before expanding: examine rework, delayed decisions, customer misunderstanding and operating burden; remove records no one uses.
Measures should follow the promise. A team could track the time spent finding current evidence, late corrections to copy and checkout, unresolved cross-functional issues, and installation dates communicated incorrectly. Even if these improve, the change cannot automatically be attributed to BDM-OS. Seasonal demand, staffing and price may change too. Measurement should inform the next operating choice, not decorate a success report. (p. 103)
An organization owns the promise.
An ad can change today. The customer experience is made by choices across many roles. When the reasons for those choices and their outcomes survive into the next project, the promise no longer depends on one person’s memory. The steps and scenario here explain the book’s proposal; they do not establish a measured business effect or an organizational model for every company.
This note adapts Chapter 9, pages 92–104, of Seoung Wi Choi’s Korean book AI 시대, 마케팅팀은 사라진다. All page numbers refer to the Korean print interior, not the English edition. Part I explains what Brand Design Management is and why it is needed.