Can you use the AI evidence brief?
Check which documents count before you tell the committee the pack is ready.
The question
A fictional internal committee wants to know whether its evidence pack meets the document-completion target. One document ID counts once. The evidence owner confirms final review status.
| This cycle | |
|---|---|
| Unique IDs due | 30 |
| Completion target in percent | 80 |
The target applies to the unique documents due at this cutoff.
The pack has 26 approved documents. The completion target is met.
“What could you say from this draft, and what would you check?”
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: Read the claim
What would you do here? Open the move you’d make.
Treat the draft rate as conditional on the count and status.
The strong move. The target compares the count of unique approved documents with the unique IDs due.
Report the target as met because the arithmetic is correct.
Tempting, but it misses something. The division checks the draft values. It leaves document identity and final review status unchecked.
See how a strong candidate says it
“26 out of 30 is 86.7% if all 26 are unique and approved. I’d confirm both conditions before using that figure.”
Why it works: The target compares the count of unique approved documents with the unique IDs due.
Step 2: Ask for the register
What would you do here? Open the move you’d make.
Use the expected IDs and owner-confirmed status from one cutoff.
The strong move. A matching ID and confirmed status give the count a consistent basis.
Count any document row present in the export.
Tempting, but it misses something. An export row can repeat an ID or describe an item that is still in review.
See how a strong candidate says it
“I’d ask the evidence owner to match the export to the expected IDs and confirm the approval status at the same cutoff.”
Why it works: A matching ID and confirmed status give the count a consistent basis.
Step 3: Reconcile the new extract
The evidence owner confirms that 4 of the 26 rows repeat IDs already counted. Of the remaining unique records, 3 still need final owner review. All 30 expected IDs are unchanged.
What would you do here? Open the move you’d make.
Use 19 approved of the 30 due and show the remaining gap.
The strong move. Remove the duplicate rows first. Count only the records the evidence owner confirms as approved.
Count all 22 unique records as approved.
Tempting, but it misses something. The evidence owner says three of those 22 still need review.
See how a strong candidate says it
“There are 22 unique records in the extract and 19 approved. That is 19 of 30, or 63.3%. That leaves 11 outstanding. The 80% target needs 5 more approvals.”
Why it works: Remove the duplicate rows first. Count only the records the evidence owner confirms as approved.
Step 4: Give the verified status
What would you do here? Open the move you’d make.
Report the gap and name the owner of the next count.
The strong move. The committee receives a verified count and a clear next check.
Say the documents establish the programme’s scientific conclusion.
Tempting, but it misses something. Document completion describes the pack’s review state. The relevant scientific owner interprets the evidence.
See how a strong candidate says it
“The pack has 19 approved documents out of 30, below the target. I’d give the committee the 11 open items and ask the evidence owner when the next status check will be ready.”
Why it works: The committee receives a verified count and a clear next check.
The answer, in short
- Decision
- Report 19 of 30 documents approved and name the evidence owner.
- Because
- The draft counted repeat rows and records that still need final review.
- What would change it
- The next owner-confirmed register could show that five more documents have cleared review.
Common traps
- Counting export rows as unique approved documents.
- Using a pack status as a scientific conclusion.
Check your answer
- I named the denominator.
- I asked for the source register.
- I corrected both count errors.
- I gave the committee a next check.
Last reviewed Nov 11, 2026