Augea Research · methodology

Retail Reality Methodology

Augea is building a privacy-conscious, methodology-first way to study what retail crypto users actually experience across routes, exchanges, wallets, payment rails, DCA, withdrawals, stablecoins, and DeFi paths. This page documents the methodology Augea will follow when it begins to collect retail-reality data.

Three commitments anchor the program:

User-reported data is different from measured snapshot data. Survey-derived claims require sample size, scope, and methodology — they are not measured fact.
No individual responses are public by default. Aggregate-only public outputs; PII minimization; explicit consent for any evidence upload.
Methodology comes before publication. Augea publishes the methodology before the data, and the methodology version travels with every claim.

No surveys are live yet. This page describes the methodology; collection has not begun.

Data source classes

Augea preserves a clear separation between five source classes. Source class travels with every record and every public claim. Reported data does not silently become measured fact.

Measured snapshot data
Adapter-captured / scraped / API-captured data with provenance, timestamp, and snapshot id. Independent of any user.
Modeled route data
Computed from observed inputs and stated assumptions. Bands and ranges where exact values would be false precision.
User-reported survey data
Subjective accounts of what respondents experienced. May be true, partial, mistaken, or misremembered. Aggregate-only in public outputs.
Receipt / evidence-supported data
User-reported data backed by an opt-in redacted artifact (screenshot, receipt). Higher confidence tier than self-report alone.
Manually reviewed / verified evidence
Evidence that has passed manual moderation review by an Augea curator AND meets methodology criteria. Strongest user-data tier.
Aggregate / public data
Aggregate cells published via consumer surfaces, research, datasets, or AI-readable outputs. Source class is preserved through joins.

Privacy and consent

Aggregate reporting by default. No public individual responses by default; no individual response will be published without explicit, separate consent and review.
PII minimization. Banded amounts; categorical fields where possible; no addresses or wallet addresses; no card or bank details.
Evidence upload would be opt-in. When evidence upload is offered, it will require explicit consent, separate from survey submission consent, with a one-click retract path within a documented window.
Redaction. Automated PII detection plus manual review before any public exposure. No raw receipts or screenshots are public by default.
No selling personal data. Survey data is never used for marketing identifiers and is never sold.
Delete / export path. A documented delete-and-export path will be in place before live collection. Deletes propagate downward through every layer.

Survey quality

No leading questions. Wording must not bias toward a specific answer.
Recall windows. Factual questions are bounded to a recall window (for example, "in the past 30 days").
"I don't know" options. Required on every factual question. Removing this option fabricates answers.
Amount bands where privacy matters. Banded ranges (for example, "$50–$200") instead of exact figures where banding does not lose meaning.
Duplicate and abuse controls. Rate-limiting, plausibility checks, response clustering, and moderation review for evidence-supported submissions before public exposure.
Sample-size thresholds before public claims. Cells below threshold render as "Insufficient sample" or are suppressed. They do not render as zero or as a confident number.
Confidence tiers. Every survey-derived claim carries a confidence tier (LOW / MID / HIGH / INSUFFICIENT). Methodology version, sample size band, and collection window are disclosed alongside.

Right of reply

Named-exchange and comparative claims may require a right-of-reply window before publication. Severe or comparative findings need review. The right-of-reply protocol exists to ensure fair process, not to grant veto: a named entity cannot suppress a valid aggregate finding by declining to respond.

Right of reply is fair process, not veto. Aggregate findings publish with the documented response status; the named entity's response is logged and summarized alongside.
Corrections / updates path. Corrections are published as new dated entries; originals are preserved with annotation; downstream outputs propagate the correction.
Workflow status is part of the claim. Statuses such as "requested, pending", "response received", "no response", "disputed", or "withdrawn" travel with every named-entity claim where right of reply applies.

What Augea may study later

The following program areas are planned. None of them are live. Each will register as its own future program with methodology, sample-size thresholds, privacy posture, evidence handling, analysis plan, and right-of-reply requirement before any collection begins.

Planned