Evidence & Decision Standard — House of Chance

HOUSE OF CHANCE / EVIDENCE & DECISION STANDARD

Evidence before action.

House of Chance and SignalForge separate what is observed from what is inferred. A recommendation is not treated as proof that a change will increase sales, revenue, conversion, or profit.

Last reviewed: September 25, 2026

Current proof status

Method published: yes — the Evidence & Decision Standard and sample audit are public.
Offer scope verified: yes — the live SignalForge product states the current $149 scope, delivery window, limitations, and no-guarantee language.
External customer outcome proof: not currently claimed. The public sample is a House of Chance self-audit, not a client case study.

External testimonials, case studies, or outcome claims are not treated as public proof unless the underlying customer evidence and permission are separately verified.

What every SignalForge finding should contain

Observation — what was directly visible on the agreed public customer journey.
Evidence — the page, step, wording, state, or customer-facing behavior supporting the observation.
Friction hypothesis — the possible customer consequence, clearly labeled as interpretation unless it has been measured.
Confidence — how strongly the available evidence supports the diagnosis.
Recommended action — a specific change or bounded test.
Verification step — how to check whether the change was implemented correctly and whether the next evidence supports keeping it.
Limitation — what the available evidence does not establish.

How we decide whether action is justified

ACT

Use when the evidence is current and strong, important contradictions are resolved, the downside is bounded, and the action has a practical rollback path.

TEST FIRST

Use when the evidence is directionally useful but not strong enough for full commitment. Prefer the smallest reversible test that can produce better evidence.

HUMAN REVIEW

Use when the action could materially affect customers, privacy, security, legal/compliance obligations, meaningful spending, or is difficult to reverse. Evidence may be strong while the decision still requires human judgment.

HOLD

Use when evidence is stale, confidence is weak, an important contradiction remains unresolved, or the decision-critical facts cannot yet be verified.

DO NOT ACT

Use when a decision has been superseded, the proposed action is unsupported, the expected benefit depends on claims we cannot substantiate, or the credible downside is not justified by the evidence.

Before you take action

Is the evidence current? If not, reverify before changing anything.
Is the issue reproducible? One-off artifacts deserve less confidence than repeatable behavior.
What happens if we are wrong? Higher downside requires more review.
Can we reverse the action? Prefer reversible moves while confidence is still developing.
What is the cost of waiting? Urgency and permission are separate; a decision can be urgent and still require a test or human review.
What is the smallest safe test? Use the least-commitment experiment that can produce better evidence.
What would make us stop? Define rollback or kill criteria before acting.
What evidence would change the decision? Name it before the test begins.

What strengthens confidence

Current primary-source or live-system evidence.
More than one independent observation pointing to the same issue.
A clearly defined customer journey and conversion goal.
Reproducible behavior rather than a one-time artifact.
A bounded recommendation with a measurable verification step.

What weakens confidence

Old screenshots or stale store states.
Conflicting source evidence.
Assuming causation from a single visible issue.
Using traffic, replies, carts, or checkout starts as if they were revenue.
Invented benchmarks, lost-revenue estimates, testimonials, urgency, or guaranteed outcomes.

What we will not ask you to send

For the SignalForge 48-Hour Revenue Leak Audit, the standard scope uses a public customer journey. Do not send passwords, API keys, card information, MFA codes, private keys, or confidential customer records.

What this means for you

A useful audit should help you answer two different questions:

What deserves attention?
Is the evidence strong enough to act now, test first, or hold?

If the evidence cannot support a legitimate finding or action recommendation, the limitation should be disclosed instead of filling the report with guesses.