METHODOLOGY / VERSION 1

Every observation needs context.

Selected coverage

Our Mexico pilot begins with the public CostoCierto editorial index. An imported record keeps its original review date and is explicitly marked as not reverified. Inclusion does not imply approval, market leadership or a lending recommendation.

Evidence levels

Official or provider disclosures, user statements, third-party estimates and our own interpretations are kept separate. A search-volume estimate is not loan demand. App installations are not active borrowers. A query’s Mexico locale is not proof that every app metric describes Mexico.

Dates and change history

We distinguish the source period, source review date, import time and first observation of a change. A baseline creates the starting point; it is not evidence of a market change. Machine-detected differences require review before they enter a research conclusion.

Comparable costs

Rates are comparable only when product, amount, term, taxes, fees and cash flows are compatible. We retain original descriptions when normalization is not justified. Missing fields remain missing.

App and search signals

DataForSEO is the selected provider for search estimates and public app information. Provider results are checked for invalid ranges and inconsistent identities. Data is released to customers only after both editorial and usage-rights review. Similarweb is not used.

Personal data

The pilot does not collect private Facebook or WhatsApp conversations, borrower lists or personal credit files. No user research submissions are accepted in this release. Any future study needs a separate, explicit participation and privacy process.

Corrections

Published research should identify its edition, sources, review date and limitations. Corrections must preserve the earlier version and explain what changed. Provider observations alone do not justify allegations about a lender.