# Hypothesis Map

| ID | Theme | Hypothesis | Decision impact | Uncertainty | Evidence needed | Status |
| --- | --- | --- | --- | --- | --- | --- |
| H-001 | Market | Creator/influencer marketing spend is still growing and becoming a core budget line. | high | medium | Market spend reports, ad industry data, platform signals | supported |
| H-002 | User | The most valuable unmet needs are workflow uncertainty reduction: selecting creators, avoiding risk, proving ROI. | high | medium | Community pains, reviews, customer interviews | supported_directionally |
| H-003 | Competitor | Leading platforms are converging on workflow, AI, measurement, and commerce rather than pure search. | medium | medium | Product pages, launches, pricing, demos | supported |
| H-004 | Commercial | Brands will pay more for ROI accountability and execution efficiency than for another generic database. | high | high | Pricing tests, sales calls, concept tests | open |
| H-005 | Distribution | Nox can win by extending its existing search -> CRM/outreach -> monitoring -> payment/governance chain. | high | medium | Internal product fit, user workflow mapping | supported_by_internal_context |
| H-006 | Feasibility | Bounded AI agents are feasible in brief parsing, shortlist generation, risk summaries, reports, and approved drafts. | medium | medium | Prototype tests, data availability, human review results | supported_with_caution |
| H-007 | Risk | Unbounded automation will create platform, compliance, trust, and brand safety risks. | high | medium | FTC/platform rules, AI governance reports, user trust tests | supported |

## Priority Questions

1. Which P0 opportunity creates measurable user value fastest: shortlist, risk, or ROI?
2. Which commerce data sources are actually accessible for Nox target customers?
3. What level of automation do users trust before requiring manual approval?

## Kill Criteria

- Users do not prefer AI-generated shortlist over normal search results in real campaign tasks.
- Creator risk summaries cannot be explained with evidence.
- Commerce attribution cannot connect to real platform/link/code/order data and becomes vanity reporting only.
- Users or sales perceive the product as “just AI search” rather than a broader operating workflow.