The Horizontal StrategyDocs Menu

Concepts Draft

Why a gate never averages

A business can score well overall and still rest on one fatal assumption. The gate reads assumptions one at a time so that it cannot hide.

Updated

The decision it serves

You have a scorecard, and it looks good overall. This idea stops a good average from approving a plan with a fatal hole in it, because averages are how fatal holes hide inside good numbers. The gate exists to make hiding impossible.

What the gate reads

The gate reads your assumptions against the bar the commitment set, and it reads them one at a time. The rules are short.

  • No load-bearing assumption below the bar.
  • No load-bearing assumption contradicted by the evidence.
  • The frame complete, with its thresholds written in advance.

That is the whole mechanism, and it is why a business can score well overall and still fail the gate. Eight strong areas and one wrong load-bearing belief is a strong-looking business that does not work.

Why an average is the wrong instrument

An average treats every claim as interchangeable, and load-bearing claims are not interchangeable with supporting ones. If the customer will not pay, it does not matter how well the operations area scored. Averaging a level-five delivery claim with a level-one demand claim produces a level three that describes nothing real. The canvas explicitly refuses to produce an overall score, and so does the interactive tool.

Below the bar is a gap. Contradicted is an answer.

The gate distinguishes two states that a scorecard blurs. An assumption below the bar is a gap, and a named test could close it. An assumption contradicted by evidence is an answer, because the evidence has already spoken. Treating a contradicted assumption as a gap to fill is one of the canvas's listed failure patterns, because it turns an answer into a chore and keeps the idea alive on paper.

Immune to the room

The gate is deliberately immune to how the room feels. That immunity is the feature the most senior person in the room will like least. It is also the protection the founder needs most, because the person most likely to skip the frame is the one who already believes.

The verdict names its rule

A verdict here is not a score. Nor is it a blessing. A proceed names its conditions, its limits and its review date, and a do-not-proceed names its route: test further, narrow, reframe, pause with a written trigger, or stop. Every verdict names the rule that decided it, so you are never left arguing with a feeling, and nobody has to remember why.

The tool records its verdict and yours separately. You can overrule it, and sometimes you should, because the canvas reads claims and you read the world. The difference is written down with your reasoning, and it goes into the Validation Decision Report.

At Halden Systems

The original whole-base Build was blocked, and the gate named the rule: two load-bearing assumptions below a Build bar. The board pack recorded the route, narrow, and the rule beside it, on a single page. The board read one page, asked two questions, and approved a narrowed pilot in eleven minutes. Nobody discussed whether the idea was good. There was nothing left for that question to do.

Failure patterns

  • An overall score computed anyway, because averages are comforting and the gate is not.
  • The system overruled silently, so the instrument takes the blame for the human decision.
  • Routes softened into scheduling language until nobody can tell a pause from a stop.
  • "We are deprioritising", which is usually pause without a trigger or stop without the courage.

Where to see it in the estate

  • The Validation Canvas, modules 06 and 07.
  • The public canvas page, in the section "No average score".
  • The Validation Decision Report the interactive tool prints.