Product
An instrument for composite indicators
One data layer, four instruments. Browse the world's indicator hierarchies, compose them into direction-aware indices, monitor the result against benchmarks that are computed, never asserted, and pull any of it, your own indicators included, over a read-only API. Alongside it, a second plane carries trade and tariff data: applied schedules from eighteen customs authorities, bilateral flows and the policy actions that move them.
Explorer
Every series, in its own hierarchy
Work the way the sources are organized: World Bank categories, UN goals → targets → series → dimensions, IMF macro forecasts, and labour, demographic, poverty, health, education, climate, energy, water, governance, conflict and innovation series from the ILO, FAO, UN agencies, WRI, Ember, V-Dem, UCDP, WIPO and more, with your own published custom indicators side by side. Pick a country, see the history, read the definition, then send it onward.
- Historical charts with units, definitions and periodicity for 18,000+ series across 22 sources, from 1750 to 2050 projections.
- Frequency disclosed: monthly or quarterly source data is annualized by a stated rule (average, sum or end of period), never silently.
- Add to portfolio from the detail view: the holding lands at the country in view, weights rescale, and the change is logged.
Series detail: metadata that travels with the number
| Series | Frequency | Latest |
|---|---|---|
| Access to electricity, % of population | Annual | 99.4 · 2023 |
| Unemployment rate, national estimate | Monthly · annualized (average) | 5.4 · 2024 |
| Renewable energy consumption, % of final | Annual | 21.8 · 2023 |
| Exports of goods, current US$ | Quarterly · annualized (sum) | 412.6B · 2024 |
Portfolios
A portfolio is a composite index
Holdings are indicator × country pairs, each carrying a weight and an orientation: which direction counts as improvement. Normalization puts hectares and kilograms of CO₂ on one unitless scale before any weight applies, so "50% of carbon intensity" means half the say in the index's movement, not half of a raw number.
- Three normalization methods: index to base year (default), z-score and min–max, with an automatic, disclosed fallback for series that cross zero.
- Five-step wizard with live construction preview from real series and a 35% concentration guardrail.
- What-if rebalancing: drag weights and the entire history recomputes instantly; apply to log the change to the activity trail.
Holdings: weight, direction, contribution Σ 100%
| Holding | Orient | Weight | Contrib (pts) |
|---|---|---|---|
| Access to electricity · BGD | ↑ better | 18% | +18.9 |
| Access to electricity · IND | ↑ better | 14% | +12.4 |
| Energy intensity · BGD | ↓ better | 12% | +6.8 |
| PM2.5 exposure · PAK 2021 vintage | ↓ better | 11% | −2.1 |
Benchmarks & attribution
Benchmarks you can compute, gaps you can decompose
A benchmark here is never a text label. Track the same basket against one comparator country, give every holding its own peer in a blended basket (the multi-asset way to isolate selection), or benchmark against a custom indicator.
- Blended peer baskets hold weights identical on both sides, so the gap is pure selection.
- Gap, hit rate and crossings computed over the common axis; crossovers raise alerts.
- Attribution waterfall regroups losslessly by holding, country or theme over any window.
Attribution since 2019: running decomposition
| Access to electricity · BGD | +18.9 | |
| Electric power consumption · VNM | +11.2 | |
| Renewables share · IND | +7.6 | |
| PM2.5 exposure · PAK | −2.1 | |
| Net composite change | +35.6 |
Data health
Estimation you can see
Development data has gaps, and pretending otherwise is how composite indices lose credibility. The engine interpolates interior gaps and carries tails forward, and tells you, cell by cell, that it did.
- Coverage heatmap: observed, interpolated, carried-forward and missing cells per holding-year.
- Vintage table with per-holding freshness; stale series are flagged when they lag the freshest data by two years or more.
- Alerts with materiality: a new release only alerts when it moves your composite.
Cell provenance: nothing silent
| Holding | Frequency | Latest vintage | Status |
|---|---|---|---|
| Access to electricity · BGD | Annual | 2024 | Fresh |
| Renewables share · IND | Annual | 2023 | Lagging 1y |
| PM2.5 exposure · PAK | Annual | 2021 | Stale · carried forward |
| Unemployment rate · VNM | Quarterly · avg | 2024 | Fresh |
Indicator Builder
When the indicator you need doesn't exist yet
Combine series from any connected source into a custom indicator: sums, differences, products and ratios, with weights, optional normalization, and per-operand frequency control for sub-annual sources. Publish it and it becomes a first-class citizen: chartable in the Explorer, holdable in portfolios, usable as a benchmark, retrievable over the API.
- Four operators, applied top to bottom: add, subtract, multiply and divide, so intensity and per-unit ratios are one formula away. Points that would divide by zero are dropped, never charted.
- Per-operand frequency: annualize by average, sum or end of period, or keep the source's native quarterly and monthly periods on the axis.
- Missing-data policy per indicator: skip years lacking any operand, or carry each operand's latest value forward.
A custom indicator, as stored and as computed
| = (Renewable electricity output × 2) | WB | avg |
| + (Energy intensity of GDP × 1) | WB | avg |
| ÷ (CO₂ per capita × 1) | UN | native |
Settings: normalization none · missing data skip years · status published
Data API
The same numbers, over the wire
A read-only, versioned REST API over the unified data layer, for notebooks, pipelines and dashboards that live outside the app. Everything it serves follows the same disclosure rules: annualization is stated, native frequency is available on request, and your published custom indicators compute on demand, identical to the chart in the app. Available on the Pro plan and above.
- Catalog, countries and observations: search the indicator catalog, then pull any series for any country by ISO code, with year-range filters.
- Your composites, computed: published custom indicators return their calculated series, operators, weights and frequency rules applied server-side.
- Keys you control: created in the app, shown once, stored only as hashes, revocable at any time. A downgraded plan pauses access without deleting anything.
Read-only endpoints, key-authenticated X-API-Key
| Endpoint | Returns |
|---|---|
| GET /v1/indicators?search=… | Catalog search with units and codes |
| GET /v1/countries | Countries with ISO alpha-3 codes |
| GET /v1/data?indicator=…&country=… | Observations, annualized or native |
| GET /v1/custom-indicators/{id}/data | Your composite, computed on demand |
Trade & tariff data
Every schedule from its own customs authority
Alongside the indicator layer, Indicator Gateway publishes a trade and tariff data plane built entirely from official national sources: applied tariff schedules, bilateral trade flows and the documented tariff actions that change them. Each jurisdiction is served at its own legal grain, in its own language, from its own publication. We never merge two countries' schedules into a single invented number, and we never fill a gap we cannot source.
- Eighteen official schedules: the United States, European Union, United Kingdom, Northern Ireland (kept separate from Great Britain, as the protocol requires), Canada, Australia, Japan, South Africa, Türkiye, Indonesia, Brazil, Argentina, China, Mexico, Korea, India, the Eurasian Economic Union and Saudi Arabia, each parsed from the authority's own gazette, workbook or portal.
- One comparable rate when you need it: a derived HS6 layer puts the applied rate for a six-digit product side by side across eighteen jurisdictions, with the method versioned in every row and non-convertible duties counted and excluded rather than guessed.
- Bilateral trade flows: monthly imports and exports by product and partner from eleven reporting authorities, bitemporal so you can replay what was known on any past date. Where a country does not publish redistributable statistics, its flows are reconstructed from its partners' reports and flagged as mirror data, never mixed in silently.
- Tariff policy as evidence: the United States tariff actions of 2025 and 2026 as structured events, each linked to the official document that created it, with effective windows, measures and rate components extracted and labelled by review status.
One product, eighteen applied rates HS 8507.60
| Jurisdiction | Applied rate | Basis |
|---|---|---|
| United States | 3.4% | HTS |
| European Union | 2.7% | TARIC |
| United Kingdom | 2.0% | UK Tariff |
| India | 20.0% | Customs Tariff |
| Brazil | 18.0% | Mercosur TEC |
| Indonesia | 10.0% | BTKI |
| Saudi Arabia | 5.0% | ZATCA |
| Japan | 0.0% | Customs Tariff |
Coverage, stated
An empty answer that tells you why
Trade data is mostly interesting where it is incomplete: confidential cells, provisional months, revisions, jurisdictions that publish nothing. The usual practice is to return a zero and let you find out later. Every response here carries its own coverage statement instead, so an empty result is only ever a true zero when the platform says the query was covered.
- Confidential values stay null, never zero: a masked cell is flagged as masked, with the provider's own secrecy band where one is published.
- Standing disclosures travel with the data: each dataset publishes its known gaps in the same response, so the limits are in front of you rather than in a footnote.
- Immutable revisions: every schedule edition and data release is a pinnable version, addressable by publication date or by legal validity, so a chart built last quarter still reproduces exactly.
- Provenance to the byte: the source document behind every figure is archived with a checksum, and every dataset states its licence and required attribution.
Coverage and freshness for every dataset are public, before you sign up or pay: see data status, the API documentation and the full licence roster.
Disclosure that rides with the row flags
| Signal | What it tells you |
|---|---|
| covered | An empty result is a genuine zero |
| not_covered | Outside what this dataset claims |
| masked_value | Trade exists, magnitude confidential |
| mirror | Reconstructed from the partner's report |
| provisional | Flash estimate, revision expected |
See it with your own mandate.
Sign up free and compose your first index against live data.