B2B Proposal case

Status: instrumented, not executed. The demo supports definition and audit of the decision; no registered action, later outcome or attributable effect exists.

The decision

Where should we intervene first to recover conversion without unnecessarily increasing acquisition cost?

The commercial demo uses simulated data and shows deterioration concentrated in Proposal plus the series' worst monthly compliance in June 2026. The first option to test is a reversible intervention in proposal quality and speed. It has not been chosen or executed.

View the primary dashboard Review the contract Audit the evidence

What we know about the simulated demo

Evidence	Reproduced result	Class	Decision strength
June 2026 weighted compliance	80.11%	Calculation	High within simulated dataset
Sellers below 90% in June	8 of 8	Calculation	High within simulated dataset
Proposal conversion, Jan–Mar	48.03%	Calculation	Provisional baseline
Proposal conversion, Apr–Jun	36.53%	Calculation	−11.50 pp deterioration
Event / referral / inbound win rate	8.93% / 7.94% / 3.60%	Calculation	Insufficient for superiority; wide overlapping intervals

What we do not know

Whether the signal exists in a real operation; whether an owner accepts intervention; expected improvement; comparability of intervention and control; and observed result. These gaps block impact claims.

DATA → IDEA → DECISION

Layer	Calibrated reading
DATA	June reaches 80.11% of target and Proposal falls 11.50 pp between observed 2026 quarters.
IDEA	Deterioration appears more concentrated in conversion than volume; Proposal intervention may be more direct and reversible than buying demand. This is inference, not established causality.
DECISION	Preregister a bounded Option A test with owner, expected range and comparison. Without them, do not execute a pilot claim.

PULSE cycle status

flowchart TD
    A["Catalogued demo"] --> B["Selected case"]
    B --> C["Instrumented contract"]
    C --> D{"Owner + real data + counterfactual"}
    D -->|No| E["Blocked for impact"]
    D -->|Yes| F["Decide and act"]
    F --> G["Observe 2–3 cycles"]
    G --> H["Compare and learn"]

v0.5.0-rc.1 reaches Instrumented contract and does not jump to result.

What the BoK adds

It adds real alternatives including no action; separation of fact, calculation, inference and hypothesis; a frozen pre-action expectation; process-quality assessment without outcome; counterfactual and attribution limits; and a longitudinal record.

Cycle 0: one recommendation changed

The source described event and referral as significantly superior to digital. Reproduction found higher rates but wide overlapping Wilson intervals. Therefore B and C are blocked, A keeps provisional preference and D remains comparator. Claims discipline corrected an internal recommendation; this does not demonstrate commercial impact or PULSE superiority.

Next gate

The case advances only after owner, selected option, expectation, group assignment and at least two cycles. See the evidence protocol and cycle record.