Read US inflation from the original CPI index
Distinguish the US CPI-U index from its monthly and yearly percentage changes. Check the original monthly endpoints, source gaps and unknown release clock before preserving an analysis.
1. Identify the index and its unit
The US inflation view uses the original BLS series CUUR0000SA0: Consumer Price Index for All Urban Consumers, US city average, all items, not seasonally adjusted. Its reference base is 1982–84=100. This is a monthly index level, not a percentage, core CPI, chained CPI or a measure of every household’s own spending.
The prepared actual overview loads on opening. Read the selected monthly range and source before interpreting a value. The base of 100 represents the average reference-period price level; it is not an annual inflation target. The separate European HICP view uses another statistical definition and source, so the two indices are not interchangeable.
2. Compare exact calendar-month endpoints
Monthly change, MoM, is (current index / previous calendar month’s index − 1) × 100. Yearly change, YoY, uses the same calendar month one year earlier instead: (current index / index twelve months earlier − 1) × 100. YoY is a twelve-month endpoint comparison, not the average of twelve monthly rates or a comparison of annual averages.
Both changes in this view are calculated from the published, not seasonally adjusted index. In particular, NSA MoM must not be presented as the seasonally adjusted monthly headline commonly quoted in news. Percentages are derived from the exact published index strings; a rounded display does not replace those inputs, and these calculations do not claim access to unpublished BLS precision.
3. Keep a genuine source gap visible
The captured source explicitly reports October2025 as unavailable, with footnote X explaining the 2025 lapse in appropriations. The original dash becomes a missing index with its explanation. It is not zero and is not replaced by September, November, an interpolation or an annual average.
A missing endpoint also makes the corresponding derived rate unavailable. October2025 has no index or derived change; November2025 has no MoM because its previous calendar month is missing. A November-to-September comparison would span two months and cannot substitute for November MoM. Future months and unreported endpoints likewise remain unavailable, even if another displayed rate is valid.
4. Separate the measurement month from release and capture
September identifies the month measured by the index. Its UTC month boundaries organise the analysis; they are not a release timestamp. Datisto’s capture clock records when the original source response was received. The API’s optional latest marker indicates the newest source row, and its responseTime describes request duration. Neither establishes when BLS first published that monthly value.
This source response supplies no per-value release or revision clock, so release time remains explicitly unknown. BLS describes unadjusted CPI-U as final when issued but can correct errors; that general policy does not establish an original-release vintage for this capture or promise that no correction can occur. A stored analysis preserves its captured inputs rather than silently taking a later source revision.
5. Inspect the inputs and preserve one result
Open an index’s Passport for its original value, source month and footnote. For a MoM or YoY cell, open the paired-endpoint Passport: inspect the current index and the exact previous-month or previous-year index together, including their original source evidence. An unavailable endpoint remains visible as unavailable; the derived percentage does not acquire an invented monthly observation.
This is a guide to Datisto. Private features require sign-in and access to your own data. Personal models use your assumptions; availability and permitted use of source data are checked with the particular result. This is not investment advice.