Canary.Research
Validity · Public Pre-Registration Ledger

Every audit is pre-registered before it runs.

Pre-registration is a discipline borrowed from clinical research: timestamp the hypothesis BEFORE the experiment runs, so no one can edit the hypothesis after seeing the data. Canary applies the same discipline to behavioral audits. This page is the public record.

What you're looking at

Each row below represents a paid Canary audit. The pre-registration record was timestamped and stored at audit kickoff — before any synthetic persona walked the page, before any AI model ran, before any finding was generated. Client identities are anonymized at the client's request; the methodology details are not.

If you ever doubt that a Canary audit found what it claims to have found, the corresponding row here proves the hypothesis was set in advance. We can't retroactively edit a public ledger. That's the point.

Registered audits

Audit CR-DEMO-0001 · pre-registered 2026-04-27

[Anonymized direct-to-consumer Shopify supplements store]

delivered
Page Type
Homepage → Product → Cart → Checkout funnel
Hypothesis (timestamped)
The target store's homepage and checkout funnel produce conversion friction at the moments where (a) shipping cost is revealed, (b) subscription intent is set, and (c) trust signals are surfaced. Five differentiated personas walking the funnel will surface friction at one or more of these moments with consensus across at least two of the three independent AI runs.
Personas Used
Persona 01 (Researcher), Persona 02 (Trust-Anxious Buyer), Persona 03 (Subscription-Resister), Persona 04 (Mobile Scroller), Persona 05 (Bulk-Order Skeptic)
Success Criterion
≥3 findings tagged Consensus 3/3 or 2/3, with at least one tied to published behavioral research (Baymard or Nielsen Norman Group).
Result
10 of 12 findings tagged Consensus 3/3 or 2/3. Hypothesis confirmed. View the audit report →

How additions work:when a new paid audit is commissioned, its pre-registration row is added here within 24 hours of kickoff. Rows are append-only — never edited, never removed. If an audit is cancelled, the row stays with status changed to “cancelled” with a timestamp.

What this isn't: a marketing claim. It's a record. The Process section of canary-research.comsays “publish the hypothesis,” and this is where it is published. If we ever say a Canary audit confirmed a hypothesis, you can verify on this page that the hypothesis was set first.

← Back to Canary Research