Case: a solution that succeeded, then failed after a department move
Info
This case extends the shared synthetic case SYN-CS-001 (the example company "Demo Service Company A"). All organizations, people, and figures are synthetic teaching data and do not represent any real company or real outcome; it is used only to illustrate how to analyze why the same solution can fail after moving to a different department.
Demo Service Company A's customer-service knowledge assistant, after going through the full IL 201-301 process in the customer-service department, is already a successful project with clear value evidence (see Module 1-2's portfolio analysis and operating model). Management then decided to copy the same system directly to the sales department, hoping sales staff could also use it to answer customer questions. Three months later, adoption in the sales department was under 10%, and most sales staff reported that 'the answers usually don't apply.' This is not a technical failure; it is a scale-out plan that never analyzed the differences between departments.
| Dimension | Customer-service department (the conditions that made it succeed) | Sales department (what actually went wrong) |
|---|---|---|
| Data quality | Knowledge-base documents are maintained by the customer-service department; outdated and contradictory documents are flagged and excluded | Sales quotes and contract terms are frequently negotiated case by case and are not covered by the knowledge base; the AI often cites expired standard pricing |
| Process | Customer-service cases follow a standardized issue-classification and reply process that the AI output can be embedded into directly | Sales work relies heavily on case-by-case negotiation and relationships, with no standardized "issue to reply" process to embed AI output into |
| Access/permissions | Customer-service agents' existing access already covers the whole knowledge base, and the read-only integration was already tested | The sales department has stricter permission tiers for customer contracts and pricing information; the AI tool's access scope was never re-tested for the sales department |
| Adoption willingness | Customer-service agents already had training and coaching support, and understand when to hand off to a human | Sales staff received no training at all, and do not see themselves as needing a "customer-service tool" |
| Skill | Customer-service agents are used to checking standardized replies for correctness, so review cost is low | Sales staff rely on sales experience and customer relationships for judgment; they are not used to checking AI-generated text, and are reluctant to review it |
| Cost | The customer-service department already budgets for reviewer labor and tool licensing | The sales department never budgeted for any of this, and no one was assigned to review output |
Warning
Common mistake: treating evidence of success in the original department as evidence it will also succeed in a new department. IL 202's data-governance readiness, IL 301's pilot results, and the IL 202 readiness review were all scoped to the original department; a department move requires re-examining these six dimensions, not carrying over the old approval.
- Step 1
Before considering scaling a solution to a new department, compare the original and new department on each of these six dimensions. - Step 2
For every dimension with a clear gap, decide whether it must be remediated before scaling, or can be remediated while scaling. - Step 3
Turn each remediation gap into a concrete 90-day action as required in the next slide, rather than a vague line like "strengthen data and training in the sales department."