Most business reports present every number with the same confidence. A sales figure pulled from the accounting system sits next to a market-size estimate copied from a press release, and both appear in the same font, in the same table, carrying the same weight. The reader cannot tell which is solid and which is a guess, so decisions end up resting on the weakest number in the room.
We grade every finding we publish or deliver. The grade describes the evidence, not how much we like the conclusion.
The four grades
| Grade | Name | What it means |
|---|---|---|
| A | Measured | Directly observed in reliable data, with the method stated. Your own sales, costs and advertising data, or a controlled test. |
| B | Documented | Stated by a primary source: a regulation, an official statistic, platform documentation. |
| C | Reasoned | Inferred from evidence through a model, comparison or analogy, with the assumptions written down. |
| D | Practitioner view | The judgment of someone experienced, labelled as opinion. |
None of the grades is bad. A practitioner’s view is often the only evidence available on a new market, and it can be valuable. The point is to know which kind of evidence you are holding.
How to apply it
1. Write the decision first
“Should we launch in Germany next quarter?” is a decision. “Germany market analysis” is not. The decision determines which evidence matters; without it, research expands to fill the time available.
2. List the facts the decision depends on
For a market launch these might be demand, landed cost, compliance requirements, competitor prices and the time to break even. Each becomes a row in a short table.
3. Grade each fact and name its source
Next to every figure, record where it came from, when, and its grade. A date matters: a platform fee or a regulatory threshold that was correct last year may not be correct now.
4. Find the weakest critical link
Look for facts that are both critical to the decision and graded C or D. These are where a small amount of extra work pays most: a supplier quote instead of an estimate, a small test instead of an assumption.
5. State the confidence of the recommendation
A recommendation is only as strong as the weakest fact it truly depends on. Say so. “Launch, with high confidence on cost and medium confidence on demand” is more useful than a confident-sounding paragraph.
A rule of thumb. If changing a C- or D-graded number by 30% would reverse the decision, invest in upgrading that number before deciding.
Common traps
- Precision as a disguise. “The market is worth 482.6 million” is still a C if it comes from a model.
- Correlation read as cause. Products with more reviews often rank higher; that alone does not show that reviews cause the ranking.
- Stale primary sources. A regulation is grade B only in the version that is in force.
- Averaging away the problem. An average margin can hide a product that loses money on every sale.
What changes when you grade
Meetings get shorter, because disagreements turn into questions about evidence rather than opinion. Research budgets go to the facts that can change the decision. And after the decision, it becomes possible to learn: was the outcome bad luck, or did we act on a weak number we knew was weak?
Weighing a decision like this?
Discuss it with usMore research