Brand positioning: write what you give up before what you choose
In annual brand discussions, when positioning is still a slogan everyone can agree with, the result is often that the front line can't use it to turn down requests.

When a new rule is introduced, rolling out a pilot to every channel at once often means the rule fails everywhere before there is time to correct it. A more reliable approach is to test it in one setting from start to finish, then decide whether to scale it. Record the decision in the pilot documentation.

If you treat a rules pilot as a one-time rollout across every channel, you can’t build a system. For a new rule, the pilot needs to be broken down; otherwise, you won’t know where the problem lies: the rule fails at scale, with no time to correct it.
Breaking it down isn’t about adding complexity. It makes it possible to test in one setting from start to finish, then decide whether to scale, with each stage playing a distinct role.
Test it in one setting from start to finish, then decide whether to scale. If you can’t do that, the rules pilot is still just a one-time rollout across every channel.
Each layer of the rules pilot needs an interface with the next. The layer above defines what cannot be compromised, so the layer below knows what it can change.
Without these interfaces, every setting reinterprets the rules pilot. A one-time rollout across every channel slips back in through the gaps.
Rules must be actionable; examples can only illustrate. If the team only saves finished work it likes, it will return to the same problem in a new setting: the rule fails at scale, with no time to correct it.
That’s why the pilot documentation should specify conditions, not point to a single reference image.
The implementation layer needs to work for everyone. When the pilot ends, check whether the documentation still points to the principle: test it in one setting from start to finish, then decide whether to scale.
If the implementation layer serves only experienced people, the rules pilot becomes a one-time rollout for a select few all over again.
When iterating, change the rules, not the names of the principles. Owner: brand lead.
At the end of each round, return to the same question: have people in this setting actually used the rule to reject something? If you can answer, the system is still intact.
From non-negotiable principles to tools usable in practice, each layer answers one question. This keeps the rules pilot from being reduced to a one-time rollout across every channel.
This layer states what cannot be compromised in the rules pilot.
This layer turns the rules pilot into conditions that can be checked.
This layer defines what can change from one setting to another.
This layer puts the rules pilot in the hands of the people using it, and specifies exceptions.
Note: This diagram is a conceptual framework. It explains a way of thinking and is not statistical data. Source: Qingmei Future Insights.
In the new-rules process, mark the moments when the rules pilot is actually called for. Don’t start by rewriting it as a one-time rollout across every channel.
Use the pilot documentation instead of longer explanations. Keep only what supports the principle: “test it in one setting from start to finish, then decide whether to scale.”
When the pilot ends, ask only: have people in this setting actually used the rule to reject something? If you can’t answer, it isn’t a decision yet.
Brand lead. A rules pilot without an owner will be quietly changed when conflicts arise.
In annual brand discussions, when positioning is still a slogan everyone can agree with, the result is often that the front line can't use it to turn down requests.
In service process discussions, when the promise is still a sentence that is complete on the outside but carries no action on the inside, the result is often that customers don't feel it at the moments that matter.
In competitive discussions, when differentiation amounts to a comparison with only the adjectives changed, customers often cannot say what makes the brand different.