When switching tools, migrate the rules before the visuals
When switching generative tools, if migration still means having the new tool imitate the old visuals, the images may match but the rules get left behind.

When guidelines for generated content use an exclusion list that merely describes the desired feel in adjectives, everyone can agree on the words while off-brand results slip through. A more reliable approach is to make the exclusion list, rather than adjectives, the standard for release—and record that decision in the list itself.

The first thing an exclusion list must stop is the use of adjectives to describe a desired feel. When this persists in guidelines for generated content, the result is predictable: everyone agrees on the words, but off-brand outputs still get through.
Keep the scope narrow. That makes it easier to uphold the exclusion list as the standard for release.
Use the exclusion list, rather than adjectives, as the standard for release. Until that happens, the list remains little more than a description of the desired feel.
Ambiguity often enters through well-intentioned additions. In trying to accommodate more situations, teams turn the exclusion list back into a set of adjectives describing a desired feel.
Every addition makes the boundaries harder to enforce unless it addresses the underlying problem: everyone can agree on adjectives, but those words do nothing to prevent off-brand results.
An exclusion list is more useful than adjectives. State when the list must not be used, so the team understands that making it the standard for release is a practical commitment.
Keep the list short enough to read in full when the guidelines are issued.
The exclusion list exists to enforce boundaries. Ideas that fall outside those boundaries can still exist, but they do not belong in the list.
It works because the brand lead owns it and has the authority to say no.
When a boundary is crossed, go beyond correcting a single output. Return to the exclusion list and add the condition that was bypassed.
Then check again with the same question: can the list directly block a generated output that fails to meet requirements?
From non-negotiable principles to tools people can use in practice, each layer answers one question, keeping the exclusion list from slipping back into adjectives that merely describe a desired feel.
This layer defines what is non-negotiable in the exclusion list.
This layer turns the exclusion list into conditions that can be checked.
This layer defines what can change in the exclusion list across contexts.
This layer puts the exclusion list in the hands of the people using it and makes the exceptions explicit.
Note: This diagram is a conceptual framework. It explains a way of thinking and is not statistical data. Source: Qingmei Future Insights.
Identify the moments in the guidelines for generated content when the exclusion list must be applied. Do not begin by reworking adjectives that describe a desired feel.
Replace lengthy explanations with the exclusion list. Keep only what supports the principle: use the list, rather than adjectives, as the standard for release.
When issuing the guidelines, ask one question: can the list directly block a generated output that fails to meet requirements? If there is no answer, the decision has not been made.
The brand lead owns this. Without someone to maintain it, the exclusion list will be quietly rewritten when conflicts arise.
When switching generative tools, if migration still means having the new tool imitate the old visuals, the images may match but the rules get left behind.
In style discussions, when style sources remain 'collecting others' finished work as the target,' the result is often that style is forever chasing someone else's last piece.
When a prompt serves as the only brief for a generation task, the tool gets to work quickly, but the decisions behind the task go unrecorded.