Several articles in this series offer scales: a graduated bar, two thresholds, a verdict under each end. This appendix says where those numbers come from, because a graduated bar looks like a research finding and almost never is one.
Three different things, often confused
A framework is a published, dated theory. Michael Porter’s five forces, published in the Harvard Business Review in March 1979, is one. A framework says where a balance of power comes from. It gives no numbers.
A question turns the framework into something measurable. « What share of its business do you represent at that supplier? » follows from Porter, but Porter never wrote it. It is mine.
A threshold is the number that separates weak from strong. It is the most visible part of a scale, and almost always the part with no source.
The three kinds of threshold I use
Listed here from the most solid to the least, and flagged in the figure captions.
- Anchored on a fact from the file. The threshold reuses a verified figure quoted in the article. Example: 3 per cent, the share of automotive in TSMC’s 2020 sales. The fact is sourced; using it as a threshold remains my choice.
- Anchored on a common order of magnitude. The threshold matches what one ordinarily observes, without any study establishing it. Example: 40 per cent of a supplier’s business.
- Purely conventional. The threshold is chosen because it speaks, and for no other reason. Examples: ten minutes, a weekend, years and billions.
The rule I hold myself to fits in one sentence: whenever a fact from the file can serve as an anchor, it serves as one. A purely conventional threshold appears only where no fact in the article can play that part.
What these scales are not
They are not models. They predict nothing and are validated against no dataset.
They are not scores. No total can be drawn from them, and two companies do not compare by adding up their positions on these bars.
They are not standards. Nobody has adopted them, and nothing obliges you to put your thresholds where I put mine.
So what are they for
For forcing a team to write down a figure it had never written down. The value of a scale lies in the gap between two columns filled with your own data, not in the absolute position of the cursor. A company that discovers it represents 3 per cent at its critical supplier has learned something, whether the comparison threshold sits at 40 per cent or at 35.
If you judge a threshold badly placed, move it and keep the question. The question is the useful part.
Where these scales appear
Part one, « Can a product become a market leader without a government’s help? »: a grid of six channels scored from 0 to 2. The scoring rule is written in the article; the split into six channels and two rows is mine.
Part two, « Leading products write their own laws »: a five-rung scale of position inside a standards body. The five rungs come from the narrative itself, where one man occupies them in turn. Putting them in order is mine.
Part five, « A public preference is not a market, it is a countdown »: no scale and no numerical threshold. Its three questions are answered by a date, a count and a document, all three taken from the reader’s own records. It is the only part of the series that proposes no number of my own making.
This list grows as the parts are published. If a published scale is missing from it, that is an oversight, and it can be reported through the site’s contact address.
