Methodology & transparency

AffiliateLedger is a neutral change-intelligence system operated by Innovate Mind Labs. It monitors public affiliate-program information, preserves historical terms, detects meaningful changes, and shows the evidence behind every claim. It is not a blacklist, a rating, or a complaint service. Positive and negative changes are treated identically: we report what changed, not whether a change is good or bad.

What we publish, and how sure we are of it

Every item in the product is one of five kinds, and we label which:

• 1. Observed fact — the raw captured page (a snapshot), with its source URL, capture time, and SHA-256 hash. This is immutable evidence.
• 2. Extracted value — a structured term (e.g. a RevShare percentage) parsed from a snapshot, shown with the exact quote it came from. Extracted values are marked pending until a human confirms them.
• 3. Calculated difference — a diff between two versions. It is a pure function of the evidence and recomputes on demand.
• 4. Interpretation — any human classification or summary of a change. Always labelled as interpretation.
• 5. Unverified claim — anything not backed by a snapshot. We publish none of these in the current release.

No published factual claim rests on AI alone. Automated extraction is deterministic and every value cites a verbatim span in the source; where a quote cannot be matched in the evidence, the value is flagged and not published as fact. AI assistance is disabled in this release.

Where the data comes from

• - Live monitoring of programs' own public pages, using an identifying user-agent, honouring robots.txt, and rate-limiting our requests. We capture only public pages and never bypass authentication or paywalls.
• - Historical backfill from the Internet Archive's Wayback Machine via its public CDX API, which gives genuine past versions of monitored pages.

We store full snapshots privately as evidence and publish extracted facts, quoted citations, diffs, and links back to the original or archived page. We do not wholesale-republish operators' copyrighted terms; we build our own database of facts and changes.

What this release deliberately does **not** do

• - No Integrity Score or rating (we will not publish a score without a defensible, published methodology).
• - No complaint, reputation, or accusation system.
• - No user-submitted allegations.

Corrections & right of reply

If a program believes a captured fact or diff is inaccurate, we record the objection against the evidence and correct verifiable errors. The immutable snapshot chain means every correction is itself auditable.

This methodology is versioned; each change to it is recorded with its own hash.