GTM Flight Recorder

Public demo · read-only

Operating company: RelayBridge (fictional)

Insights

At a glance

Observed 90-day opportunity rate, all decisions

38.9% observed (58 of 149 eligible decisions; n = 149)

Cohort total 289 decisions at ledger cutoff 1597. Observed on synthetic data: descriptive, not causal, and not a conversion probability. How to read it, every exclusion and every denominator follow below.

How to read this page

Population

The population is every account-prioritization decision recorded by the cutoff for the accounts listed on this site. Every count below comes from the effective outcome versions and their effective attribution results at that cutoff, never from raw row counts.

Ledger cutoff
1597
Decisions in population
289
Reconstructed exactly
289
Reconstruction failures
0
Outcome window
90 days
Comparison workflow cohort
v4.1

Observed 90-day opportunity rate, all decisions

Observed rate
38.9% observed (58 of 149 eligible decisions; n = 149)
Cohort total
289
Eligible decisions
149
Positive decisions
58
Excluded: other evaluation period only
24
Excluded: reconstruction failed
0
Excluded: not evaluated
116

A decision counts as positive when it holds at least one recorded positive observation in a completed 90-day evaluation window, counted once per decision. The rate is not a conversion probability. The three exclusion counts are reasons that may overlap, so they do not add up to the cohort total.

Coverage

Effective outcome observations

Observations that are unresolved or awaiting attribution are counted here and enter no decision-level count or rate on this page. An open observation is never a completed outcome, whatever its recorded values. A closed observation with an unknown opportunity value is never a negative.

awaiting attribution
2
unresolved
24
direct
207
inferred (heuristic)
26
open window
54
closed known
195
closed unknown
10
Other evaluation period
36
Qualifying 90-day observations
149
Total effective observations
259

Decisions by observation standing

evaluated
173
unknown
9
open
51
unattributed
56
Total
289
Eligible for the 90-day rate
149

Decisions counted as evaluated and decisions eligible for the rate differ by design: an evaluated decision may hold only observations of another evaluation period, or may have failed reconstruction. A decision with no attributed observation is unattributed, including a decision with no action recorded for this decision.

Reconstruction failures

No decision failed exact reconstruction at this cutoff.

Signals (demonstration checks)

The six states are exclusive and add up to the population; only known true and known false form the comparison arms. Input available counts decisions whose sealed context held the key, input consumed counts those whose recorded decision consumed it, and an available-but-ignored input is never counted as consumed.
Signal Predicate Comparison signal present (known true) known false unavailable absent not applicable reconstruction failed Input available Input consumed
recently_funded funding_event observed within 90 days before the decision boundary 97 121 29 42 0 0 218 218
verified_integration_pressure_high verified_integration_pressure equals "HIGH" 77 167 14 31 0 0 244 57

Signal recently_funded

Historical rule matched, by the decision's own logic artifact
Logic versionArtifact hashRuleDecisionsMatched
v3.2 db3a8bdebf2befe286ab49a2381dfe6fb931ac6f848923d35e0e732adcc82db0 funding_event observed within 90 days before the decision boundary 222 91
v5.1 b5a6f33dd592fe804d9b59e081639c6527a9c4ca2727cb6f850c6888b3ea7a3b funding_event observed within 90 days before the decision boundary 67 6
Observed 90-day opportunity rate by signal state; the three exclusion counts are reasons that may overlap
Arm Cohort total Eligible Positives Observed rate Excluded: other period only Excluded: reconstruction failed Excluded: not evaluated
known true 97 58 23 39.7% observed (23 of 58 eligible decisions; n = 58) 9 0 30
known false 121 55 21 38.2% observed (21 of 55 eligible decisions; n = 55) 8 0 58
Difference
+1.5 percentage points

Signal verified_integration_pressure_high

Historical rule matched, by the decision's own logic artifact
Logic versionArtifact hashRuleDecisionsMatched
v3.2 db3a8bdebf2befe286ab49a2381dfe6fb931ac6f848923d35e0e732adcc82db0 not applicable 222 0
v5.1 b5a6f33dd592fe804d9b59e081639c6527a9c4ca2727cb6f850c6888b3ea7a3b verified_integration_pressure equals 'LOW' 67 16
Observed 90-day opportunity rate by signal state; the three exclusion counts are reasons that may overlap
Arm Cohort total Eligible Positives Observed rate Excluded: other period only Excluded: reconstruction failed Excluded: not evaluated
known true 77 41 25 61.0% observed (25 of 41 eligible decisions; n = 41) 4 0 32
known false 167 88 25 28.4% observed (25 of 88 eligible decisions; n = 88) 16 0 63
Difference
+32.6 percentage points

Workflow versions

Decisions and the observed 90-day opportunity rate by workflow version; the three exclusion counts are reasons that may overlap
Workflow version Decisions evaluated unknown open unattributed Eligible Positives Observed rate Excluded: other period only Excluded: reconstruction failed Excluded: not evaluated
v4.1 158 102 5 25 26 91 43 47.3% observed (43 of 91 eligible decisions; n = 91) 11 0 56
v4.2 131 71 4 26 30 58 15 25.9% observed (15 of 58 eligible decisions; n = 58) 13 0 60

Workflow comparison

Compared
v4.2
Against
v4.1
Observed rate, v4.2
25.9% observed (15 of 58 eligible decisions; n = 58)
Observed rate, v4.1
47.3% observed (43 of 91 eligible decisions; n = 91)
Difference
-21.4 percentage points

Each arm's cohort total, eligible count, positives and exclusion counts are in the workflow table row of that version.