Which pilot build can the lead advance?
Bramble Devices tested 30 pilot units in each of three desk sensor variants.
The question
The release lead can advance variants independently. Each needs at least 95% passing units and a complete result for the intended firmware and assembly configuration.
| Units | |
|---|---|
| Variant A | 30 |
| Variant B | 30 |
| Variant C | 30 |
Use each variant’s own cohort. Each unit keeps its original assembly variant and serial number.
“Recommend which variants can advance and identify the checks that would change the decision.”
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: How many passes does each variant need?
What would you do here? Open the move you’d make.
Use 29 passes per variant and check its configuration record
The strong move. This applies both parts of the release criterion.
Use 28 passes because 28.5 rounds down for a count
Tempting, but it misses something. 28 of 30 is below 95%.
Check only the combined rate for the 90 units
Tempting, but it misses something. The case requires each variant to meet its own threshold.
See how a strong candidate says it
“Each 30-unit cohort needs at least 29 passes and a complete result for its intended configuration. I would ask for the records by variant and serial number.”
Why it works: 95% of 30 is 28.5. The minimum count is the next whole unit.
Step 2: What does the first report support?
The first report lists 30 passes for A, 29 for B and 28 for C. It has not been checked against the firmware log.
What would you do here? Open the move you’d make.
Advance all three because 96.7% is above 95%
Tempting, but it misses something. C misses the requirement for its own cohort.
Treat A and B as provisionally eligible and check their records
The strong move. The counts fit for those two. Configuration still needs confirmation.
Hold all three until C passes
Tempting, but it misses something. The release lead can advance each qualifying variant independently.
See how a strong candidate says it
“The first report gives 87 of 90 passes, or 96.7%. A and B appear to meet the count. C has 28 of 30, which is below the minimum. I would check the serial-level configuration record before I recommend release.”
Why it works: The combined rate can pass even when one variant misses its own threshold.
Step 3: Which variant still has complete evidence?
The build log shows that 2 of B’s reported passing units ran earlier firmware. Both serials still belong to B. The log confirms the intended configuration for the other 88 units.
What would you do here? Open the move you’d make.
Advance A and B because B had 29 first-report passes
Tempting, but it misses something. Two of those B passes use earlier firmware.
Drop the two B units and compute 27 of 28
Tempting, but it misses something. The units remain in B’s 30-unit cohort. Their results need verification.
Recommend A alone and keep B and C open
The strong move. A has the complete qualifying result. B needs two current results and C misses its count.
See how a strong candidate says it
“Only A meets the rule now. B has 27 confirmed passes, 1 fail and 2 unverified results from 30. Both unverified units need to pass in the intended configuration for B to reach 29. C remains below the threshold. The corrected total is 85 of 90 confirmed passes, or 94.4%.”
Why it works: A result from earlier firmware cannot count for the current B configuration. The two units stay in B’s denominator.
Step 4: What decision do you give the lead?
What would you do here? Open the move you’d make.
Advance A and assign separate follow-ups for B and C
The strong move. The recommendation matches the evidence and keeps the open issues visible.
Wait for B before advancing A
Tempting, but it misses something. A meets both conditions and the variants can advance independently.
Report 94.4% overall and leave the variant actions to the next shift
Tempting, but it misses something. The aggregate does not identify the eligible variant or the two open actions.
See how a strong candidate says it
“I would recommend advancing A alone. I would ask test to verify both B units on the intended configuration and ask C’s variant owner to investigate its misses. I would retain the first report with the correction.”
Why it works: Each team gets a next action, and the release lead can see why the recommendation changed.
The answer, in short
- Decision
- Recommend advancing A alone.
- Because
- A meets the count and configuration requirements. B has two unverified units and C has too few passes.
- What would change it
- B can meet its threshold if both units pass in the intended configuration.
Common traps
- Keep each unit in its original cohort.
- Check the result against the intended firmware.
Check your answer
- I found the count each variant needs.
- I kept the first report provisional.
- I corrected the firmware evidence.
- I assigned two distinct follow-ups.
Last reviewed Nov 11, 2026