---
title: Methodology and sources
nav: Methodology
description: Where the economic data comes from, how coverage and freshness are handled, and which markets cannot be analyzed.
group: trust
order: 2
status: shipped
updated: 2026-09-21
---

A recommendation is only as good as the data under it, so the data is stated rather than implied.

## Sources

Economic observations come from published international sources — principally World Bank indicators for purchasing-power parity, price level, and GDP per capita, with IMF data used to fill gaps, and a separate exchange-rate provider for FX.

Observations are stored immutably with the date they refer to. A recommendation references the observations it used, so a target calculated last month can be reconstructed from the data as it stood then.

## Coverage

The canonical registry covers 191 markets. 190 carry PPP, GDP per capita, and an exchange rate.

Coverage is not decoration: it is 35% of the confidence score. A market with only PPP scores a third on the coverage component, which alone costs it roughly 0.23 of its total confidence.

## Billing currency, not country currency

Some markets are billed by the store in a currency other than the country's own — 73 of 173 in one representative catalogue.

These are analyzed and denominated in the **billing currency**. Dropping them as currency mismatches would silently remove a large fraction of a catalogue from analysis, which is worse than analyzing them correctly in the currency the customer is actually charged.

## Markets that cannot be analyzed

Five markets have no usable indicators:

| Market        | Reason                                            |
| ------------- | ------------------------------------------------- |
| Taiwan        | Not a World Bank member. Exchange-rate data only. |
| Gibraltar     | No indicators published.                          |
| Liechtenstein | No indicators published.                          |
| Monaco        | No indicators published.                          |
| Vatican       | No indicators published.                          |

These should read as explicitly unanalyzable rather than as missing data. A known gap and an unexplained blank are different conditions and should not look the same on screen — presenting them distinctly is open work.

## Markets where the data is wrong

Dollarized economies can produce a PPP ratio that is arithmetic rather than economics, pinning the market to a strategy bound. Zimbabwe and Somalia both behave this way and should be treated as [protected markets](/guidelines/protected-markets).

A market sitting exactly on a bound is a prompt to inspect the inputs, not a recommendation to act.

## Freshness

Freshness carries 20% of the confidence score. Economic indicators are published with a lag measured in months to years; PPP figures in particular are revised.

The system does not pretend to real-time economic data, and 177-country coverage is a statement about breadth, not about latency.

## Reproducibility

Every calculation stores a snapshot: the raw and adjusted economic factors, the FX rate, the candidate, constrained, and rounded prices, whether an override applied, whether the market was protected, the rounding increment, and the confidence score.

Any target can be taken apart after the fact. That is the property the whole product rests on — see [Formulas](/guidelines/pricing-formulas) for the exact arithmetic.

## What the method does not do

It does not model demand, elasticity, or competitive position, and it does not observe what your customers did. It derives a defensible starting position from published economic data. Measuring outcomes is [planned, not built](/guidelines/revenue-signals).