How Your Stats Work

1. Supported data and market

Use original Uber Eats order CSV files or ZIP archives containing Eats/user_orders CSVs. Orders in Canadian dollars (CAD) and US dollars (USD) are supported, one currency at a time; if an export has both, you choose which to replay. Other currencies are excluded, not converted; this is not geolocation or proof of a Canadian order. The app reports coverage based on the supplied records and requires at least three completed orders in the chosen scope.

2. Reconstructing and counting orders

Where a unique order identifier is unavailable, the app groups rows using request time, restaurant, city and currency. Each reconstructed order’s recorded total is counted once, even if repeated across item rows. Matching groups repeated across exports are reconciled. Conflicting versions stop the import instead of silently choosing one.

Identical item rows within one export remain separate because they may be legitimate quantities. Two genuinely distinct orders sharing every grouping field could be indistinguishable. The source export—not an independent account reconciliation—sets what can be known.

3. Spending and quantities

Only completed orders in scope contribute to recap spending and habits. Negative recorded totals remain signed amounts, not confirmed refunds. Unlisted tips, taxes, fees and adjustments are not guessed. Recorded customization amounts are not added again to the overall order total.

Orders containing an item and total item units are different metrics. Missing quantities are unknown, not assumed to be one. Restaurant listings and exact item names remain distinct; differently named branches or dishes may not be merged.

4. Dates, years and frequency

Hours and days follow the exported local-time field without a second timezone conversion. “First order” means first in your selected records, not necessarily first ever. Annual charts can include other years within the same city and currency scope. A calendar-year choice does not demonstrate complete coverage.

A streak counts consecutive dates with completed orders. A break counts full days between recorded orders; missing data can create apparent breaks. Midnight orders are those from 00:00 through 04:59. Frequency uses the observed date span and order count, not proof that you continuously used Uber during that span.

5. Personality, wording and equivalents

Rules select eligible chapters and a personality family using your own order patterns. A local session seed varies wording; it is not an identifier and is not transmitted. The same scope stays consistent within a session, while another session may use different wording for the same facts. Limited data receives fewer chapters.

Labels are editorial choices, not scientific classifications, comparisons against other customers, or inferences about age, wealth, health or household size. Spending-equivalent budgets are illustrative divisions of a recorded total, not predictions or guaranteed savings.

6. Review and report errors

Check city, year, currency, dates and the share-card preview. Keep the original export. A problem in the recap does not change your account with Uber. Report reproducible errors by email (see Security and reporting problems) with synthetic/redacted examples, not your full export. Unsupported formats should produce an error rather than fabricated results.