| name | product-impact |
| description | Translate supported defects into user, trust, conversion, support, operational, and financial impact without presenting unmeasured business loss as fact. Use after causal adjudication. |
Product Impact
Objective
Explain why a supported finding matters in the real journey, while keeping measured facts, modeled scenarios, and unknowns separate.
Impact analysis is evidence and advice. It does not grant authority to change pricing, contact a company, publish a claim, deploy a fix, or commit business resources.
Impact chain
technical or content condition
-> visible user state
-> user interpretation
-> behavior change
-> journey outcome
-> business or operational consequence
Every arrow needs evidence or an explicit hypothesis label.
User-impact lenses
Assess the applicable dimensions:
- goal completion;
- time and cognitive load;
- error prevention;
- control and reversibility;
- visibility of system status;
- accessibility and reachability;
- trust and credibility;
- privacy and perceived safety;
- recovery after failure;
- consistency across time, device, locale, and session.
Business-impact lenses
Assess:
- acquisition and first-use conversion;
- activation and task completion;
- retention and repeat use;
- support contacts and manual handling;
- operational rework;
- compliance or governance exposure;
- brand and partner trust;
- engineering opportunity cost;
- revenue or margin risk.
Evidence classes
For every impact statement, choose one:
MEASURED — supported by analytics, experiments, tickets, logs, or financial data;MODELED — calculated from explicit assumptions and ranges;QUALITATIVE — supported by the journey and human-factors reasoning, without a numeric estimate;UNKNOWN — data is missing.
Do not convert MODELED or QUALITATIVE into a factual loss number. Missing load-bearing data keeps the impact claim in NEEDS_EVIDENCE or INCOMPLETE.
Scenario model
When useful, estimate a range:
eligible users
x exposure rate
x affected-step rate
x incremental abandonment or rework
x value per completed outcome
= modeled impact range
List every assumption, use low/base/high scenarios, and identify the metric that would falsify the model.
Prioritization
Rank findings using separate dimensions:
- user harm;
- business exposure;
- evidence confidence;
- affected population;
- recurrence;
- recoverability;
- remediation cost;
- learning value.
Do not let a dramatic but weakly evidenced finding automatically outrank a smaller confirmed defect on a critical journey.
Output per finding
Return:
- affected user goal;
- visible friction or risk;
- likely behavior response;
- journey and business consequence;
- evidence class;
- assumptions and uncertainty;
- measurement plan;
- cheapest high-value experiment or fix candidate;
- success metric and guardrail metric;
- required decision owner and authority boundary.
Recommendations remain hypotheses until implementation and measurement confirm them.
