skills/audit-marketing-reasoning/references/implementation-and-freshness.md

# Implementation, Freshness, and Ethics

## Information freshness

Date the claim, source, dataset, and operating premise. Use these only as initial review heuristics, never as strict service levels:

- AI and fast-moving technology: review around 30 days.
- Platform specifications and channel tactics: review around 90 days.
- Organizational relationships and political assumptions: review around 365 days.

Shorten the interval when new citations, official statements, policy changes, competitor moves, or product releases accumulate. Lengthen it only when terminology, mechanisms, and observed behavior have stabilized. When sources conflict, require human judgment rather than averaging incompatible claims.

## Implementation audit

### People

- Who owns the decision and the outcome?
- Whose workload, authority, or evaluation changes?
- Is resistance observed, inferred, or merely possible?

### Process

- What changes in the actual workflow, including spreadsheets and unofficial workarounds?
- What approval, legal, data, procurement, or customer-support dependency exists?
- Can the organization observe leading signals before irreversible scale?

### Incentives

- Do local KPIs reward behavior that undermines the system goal?
- Does the proposal require someone to expose prior errors or surrender control?
- Is apparent certainty being rewarded more than learning?

Treat these as system hypotheses. Do not allege motives without evidence.

### Capacity and reversibility

- Can sales, operations, delivery, customer success, and measurement absorb the change?
- Can rollout be staged by region, segment, channel, or cohort?
- Define stop, continue, expand, and rollback conditions before launch.

## Ethical and trust audit

- Do not use ambiguous engagement as consent for aggressive sales contact.
- Do not design dark patterns, deception, or pressure that weakens informed choice.
- Include the experience of non-buyers, rejected candidates, churned customers, and support burden where relevant.
- Examine whether short-term KPI improvement transfers cost or harm to customers, employees, partners, or future demand.
- In behavioral-economics reviews, identify cognition, emotion, and situation together; report both expected effect and trust risk.

## Staged adoption pattern

1. Preserve the useful mechanism as a hypothesis.
2. Select the smallest reversible environment where it can be observed.
3. Align one owner, one decision, and a limited set of signals.
4. Document operating exceptions and incentive conflicts.
5. Expand only after both market evidence and delivery capacity are credible.