Method and evidence grades
Every figure in this section carries a letter after it. That letter is not a disclaimer - it tells you how far the figure can be used. Grade A can go straight into a contract. Grade B can support a judgement. Grade C can only set a direction. Using a C-grade number to sign against is the one misuse we would ask you to avoid.
The three grades
| Grade | Definition | Use it for | Do not use it for |
|---|---|---|---|
| A Official primary | Government gazettes, regulator filings, standards-body catalogues, listed-company annual reports - anything retrievable by document number | Contract terms, acceptance criteria, compliance files | - (but always confirm you hold the latest version) |
| B Authoritative secondary | Published research from consultancies, investment banks or industry bodies, traced to publisher and publication date; or vendor-published product and financial figures | Forming a judgement, setting direction, justifying a decision internally | Acceptance criteria (the underlying method and sample are usually not visible) |
| C Our own derivation | Our arithmetic, synthesis or judgement on top of A and B material; or second-hand figures we could not cross-verify | Setting direction, working out what to ask | Citing as fact, writing into contracts, using as acceptance criteria |
Three rules we hold ourselves to
- We do not reconcile conflicts - we publish them. When sources disagree on the same metric we neither average them nor pick one. We publish the spread together with the likely reason for it. The tenfold gap on the semiconductor page and the contradictory share figures on the harmonic drive page are both this rule in action. A smoothed-over number is far more dangerous than an honest range.
- For anything perishable we give the method, not the number. Lead times, prices, shares and control status all move. So we prioritise where to look and which number to look up over copying a value that expires. Every
Last verifiedline on these pages means exactly that. - If we did not verify it, we say we did not verify it. Where we could not reach the primary text, the source list says so explicitly. That is not modesty - it is telling you which step you still need to take.
Where these pages touch export controls, we publish only facts and proposal status that you can check yourself by docket number, and we separate "proposed" from "in force" explicitly. We offer no workaround, and we make no assertion about whether any specific company appears on a restricted list - an assertion of that kind, unverified against the primary source, is paid for by the reader. That is not acceptable to us.
Four steps to verify any figure yourself
- Read the grade before the number. A C-grade figure is a direction however precise it looks. Decide first whether it is worth your time to check.
- Follow the source list to the document number. Every page's sources section gives the file number, docket number or publisher. If it cannot be retrieved by number, it cannot be checked.
- Confirm the date. Check
Last verifiedat the top of the page and the update line at the bottom. Control status, lead times and market shares need rechecking beyond a quarter. - Ask the supplier for the measured value for your own specification. An industry range never substitutes for a measured report on the specific model - this is the one step on this page you should not skip.
Review cadence
| Data type | Review frequency |
|---|---|
| Export control status, tariffs, standard numbers | Before every shipment; at least quarterly |
| Lead times, prices, capacity | Quarterly |
| Market share and localisation rates | Every six months |
| Cost structure and BOM shares | When the underlying study is updated |
Related
Last updated 2026-10-09.