/answer-checks
Every answer, checked before anyone acts on it.
ContextBase checks every answer against the data behind it and what your team approved. The checks arrive with the answer, and every problem comes back named.
Not a score. Named findings, counted, on every answer.
Four checks read every answer beside the queries it ran and the rows they returned. Each problem comes back as a plain sentence.
How many members were active in March?
71,904 members were active. The count includes members whose term date falls after March 31.
Filters behave as assumedUnmet
This filter is on a nullable column, so it silently dropped rows.
4 checks ran3 met · 1 unmet
01 · What gets checked
Four checks read every answer.
Does the data returned support what was said? Is the population right? Was the right table read? Do the filters do what the answer assumes? Pick a check to see its verdict, its finding, and its spread.
Were copays applied correctly in plan year 2026?
Copays were applied correctly on 96% of claim lines in plan year 2026. The count covers members enrolled at any point in 2026 and reads the claim line table, the source of record for cost share, excluding reversed lines.
1 unmet · 3 met · 4 checks ranThe answer arrives with its checks, before anyone acts on it.
Copays were applied correctly on 96% of claim lines: Unmet
02 · How they think
Eight rules the checks keep.
The checks are honest about what they know. Three outcomes, a separate state for a check that could not run, and the weight behind every verdict.
01 / 08
No overall score
A pass and a fail could land on one number. You get counts and named findings.
Demo data
03 · How a finding is added
What one answer learns, every answer checks.
Some facts about your estate never show in a query. When your team approves one, it attaches to its table. Every later answer that reads the table carries it.
Answer · active members, March. 82,140. Enrollment report shows 87,400
Demo data
Every answer carries its receipts.
The queries that ran, the rows returned, the sources read, what was checked, and what could not be. An approved finding can carry a recipe that re-checks it against the source on a schedule.
- Act on the right answersYou know which answers to act on before anyone acts on a wrong number.
- Defensible numbersEvery answer carries its receipts, so it can be traced and defended.
- Knowledge that reaches every answerEach approved finding reaches every later answer that reads its table.
- Honest about its limitsWhich is why “no problems found” means something.
Data supports the answer · unmet“copays applied correctly on 96% of claim lines.” Facility lines were never queried, so 96% covers professional lines only.
3 queries · 41,206 rows · 2 sources · 4 checks, 1 unmet · re-checked every MondayDemo data
Four answers, checked.
Pick an answer, then open a check to see the spread, the finding, and the query behind it.
Were copays applied correctly on those members’ claims?
Do not rely on this number yet.
5 of 8 checks found something
The answer states things the data does not show.
Check every figure against what the queries returned.
The answer says copays were applied correctly. The queries counted lines that carry a copay; none compared a copay with the plan’s benefit.
SELECT s.line_type, COUNT(*) AS lines,
SUM(CASE WHEN s.copay_amt > 0 THEN 1 ELSE 0 END) AS with_copay
FROM claims.claim_line_summary s
JOIN enrollment.member_eligibility_v e ON e.member_id = s.member_id
WHERE s.plan_year = 2026
AND s.copay_waived = 0
GROUP BY s.line_type;The filter on copay_waived excludes rows. The number is probably understated.
Fix the filter and re-run. Do not use this figure as it stands.
This filter is on a nullable column, so it silently dropped rows: copay_waived = 0 excludes every line where the flag is empty.
SELECT s.line_type, COUNT(*) AS lines,
SUM(CASE WHEN s.copay_amt > 0 THEN 1 ELSE 0 END) AS with_copay
FROM claims.claim_line_summary s
JOIN enrollment.member_eligibility_v e ON e.member_id = s.member_id
WHERE s.plan_year = 2026
AND s.copay_waived = 0
GROUP BY s.line_type;The answer ignored knowledge we surfaced.
Check whether the caveat applies here.
Knowledge was surfaced, but none of it was used. Read from the recall record, which marks each surfaced item the answer used. No model involved.
The population is close but may not match the question.
Check the date range and filters against the question.
The question is about members at their out-of-pocket max. The main query reads every member’s lines, so the share could differ for that group.
SELECT s.line_type, COUNT(*) AS lines,
SUM(CASE WHEN s.copay_amt > 0 THEN 1 ELSE 0 END) AS with_copay
FROM claims.claim_line_summary s
JOIN enrollment.member_eligibility_v e ON e.member_id = s.member_id
WHERE s.plan_year = 2026
AND s.copay_waived = 0
GROUP BY s.line_type;This table is derived. It may not reconcile to line detail.
Confirm it is the right source for this question.
claim_line_summary is pre-aggregated. claim_line is the authoritative source for cost share.
SELECT s.line_type, COUNT(*) AS lines,
SUM(CASE WHEN s.copay_amt > 0 THEN 1 ELSE 0 END) AS with_copay
FROM claims.claim_line_summary s
JOIN enrollment.member_eligibility_v e ON e.member_id = s.member_id
WHERE s.plan_year = 2026
AND s.copay_waived = 0
GROUP BY s.line_type;Passed Queried the estate · Relevant knowledge surfaced · Returned rows
Not run No unsurfaced pitfalls · Re-run verification
These say what was examined about this answer. They do not say the answer is correct: a query can be wrong in a way none of them can see.
Sources read
claims.claim_line_summary318,904enrollment.member_eligibility_v88,102benefit.accumulator_ledger186,420
What teams ask about the checks
Does a check ever say an answer is correct?
No. The strongest result is “no problems found,” with the number of checks that ran. A clean result on two checks never passes for one on four.
What happens when a check cannot run?
It shows as couldn’t check, its own state. It is never counted as zero and never left silent.
Why not one confidence score?
We built one and it could not be read: a clear pass and a clear fail could land on the same number. Counts and named findings can be.
How do checks learn what our estate knows?
Through findings your team approves. Each attaches to the table it describes, every later answer that reads the table carries it, and the checks read it too.
Can a finding be re-checked later?
Yes. An approved finding can carry a recipe that re-checks it against the source on a schedule.
Start with a number your team has to defend.
Bring that number, and see what the checks find in it.