X data API company demand · trends
Daily counts of public posts on X about each company's brands, split into 16 metrics, plus a catalog of consumer-behaviour trends. Raw counts, spike-corrected counts and the editorial flag come side by side on every row.
kdate, observed_at, history_mode, release_id) work the same on every route that takes them. They are described once, in point-in-time reads. Every route also takes format (formats).Identity and coverage
GET /v1/universe
Every entity (one company's consumer business in one category) with its ids, division and ownership window. The live roster is the rows whose coverage_status is covered. Delisted and acquired names are listed too, so a universe taken at any kdate has no survivorship gap.
| Parameter | Notes |
|---|---|
ticker | Comma list of tickers, as of kdate. |
source_ticker | Comma list of source tickers (stable across renames). Use it when ticker is blank. |
company_id | Comma list of company ids. |
core_only | true (default) leaves out NONCORE entities; false includes them. |
kdate … | The roster as of a knowledge date. See point-in-time reads. |
cursor | Paging. |
Row fields, in order: entity_id, ticker, company_id, perm_id, figi, company, division, is_primary, valid_from, valid_to, then coverage_status (covered, delisted or acquired), coverage_end (equal to valid_to), coverage_start, ticker_recorded_from and the other roster fields. Subscription keys also get brand_label on carve-outs.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/universe?ticker=COIN"
{ "dataset": "universe", "rows": 2, …,
"data": [
{ "entity_id": "COIN#coinbase", "ticker": "COIN", "company_id": "6479", "perm_id": "5037999423",
"figi": "BBG00ZGF7771", "company": "Coinbase Global, Inc.", "division": "Crypto Exchanges & Miners",
"is_primary": true, "valid_from": "", "valid_to": "", "coverage_status": "covered", … },
{ "entity_id": "COIN#deribit", "ticker": "COIN", "company_id": "6479", …,
"division": "Crypto Exchanges & Miners · Deribit", "is_primary": false,
"valid_from": "2025-08-14", "valid_to": "", "coverage_status": "covered", … }
] }
Trimmed. COIN#deribit is a carve-out: its division reads "Category · Brand", it is not primary, and none of its rows are dated before its valid_from (see ownership windows).
GET /v1/resolve
Resolves exactly one identifier to its company, optionally as of a kdate. Supply exactly one of ticker, company_id, perm_id, figi or share_class_figi.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/resolve?ticker=CELH"
{ "dataset": "resolve", "rows": 1, …,
"data": [ { "company_id": "5735", "company": "Celsius Holdings, Inc.", "ticker": "CELH",
"perm_id": "4297200628", "figi": "BBG000NY37X4", "valid_from": "…", "valid_to": "…", … } ] }
Trimmed. An identifier that maps to nothing is 404; one that maps to two companies is 409. A ticker resolves only inside its dated span: a former or later ticker gets a 404 whose body lists ticker_history (each company that used it, and when). See errors. FIGIs are checked against OpenFIGI; permIDs are as supplied by the source and not independently confirmed with LSEG. External mappings are observed from 2026-09-23 onward.
GET /v1/crosswalk
Stable company ids with their verified ticker validity windows and external ids. Rows: company_id, company, ticker, source_ticker, valid_from, valid_to, mapping_status, perm_id, figi, share_class_figi, external_valid_from, external_valid_to, figi_validation, perm_id_validation.
| Parameter | Notes |
|---|---|
ticker, company_id | Optional comma-list filters. |
kdate …, cursor | As above. |
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/crosswalk?ticker=AAPL,CELH&format=csv"
GET /v1/coverage
One row per company of the current publication: is it available, from when, and through which day is it complete. Each company is listed under today's ticker, so every listed ticker works on the company routes. Trial keys get ticker and status only.
| Field | Meaning |
|---|---|
status | available, unavailable, or historical_only (ownership has ended; its last owned history is still served). |
coverage_start / coverage_end | First and last observed day of the company's series, inside its entities' ownership windows. |
complete_tail_through | The last complete tail day within complete_tail_scope; when scoped to unheld_metrics, held metrics are excluded from this completeness claim. |
refresh_status / refresh_reason | How the latest refresh treated the company, and why it was held, if it was. |
complete_rows, missing_rows, review_pending_rows | Row counts. |
company_id, source_ticker, current_ticker | Identity. |
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/coverage?ticker=CELH"
ticker also matches a company's source ticker: bebe stores is listed as BDST and ticker=BEBE returns that row. No cursor.
GET /v1/company/{companyId}/entities
A company's divisions (entities) as of kdate, with the same fields as /v1/universe. The roster is point-in-time: M&A adds and retires divisions. A delisted or acquired company still returns its roster.
| Parameter | Notes |
|---|---|
companyId (path) | The company id (e.g. 6 for Apple), or a ticker. id_type forces the namespace: company_id, ticker, perm_id, figi, share_class_figi. |
core_only | true (default) or false. |
kdate … | As above. |
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/company/6/entities"
Company series
GET /v1/company/{companyId}/metrics
Daily rows of entity × date × metric, or one company line per date × metric with rollup=ticker. /v1/company/{companyId}/demand is the same route under its older name.
| Parameter | Notes |
|---|---|
companyId (path) | Company id or ticker; id_type as above. |
metric | Comma list, or all (default). The 16 metrics: mentions, pi, npi, ps, ns, switch_in, switch_out, promo, political, price_value, price_pain, defect, outage, availability, safety, labor. Definitions in the data dictionary. |
start, end | Inclusive Eastern days. Trial ranges stop at the embargo cutoff. |
rollup | entity (default) or ticker (the company's entities summed per day). |
entity | One entity id, URL-encoded (# as %23): entity=COIN%23deribit. An unknown id returns no rows. |
core_only | true (default) or false. |
column | both (default), raw or corrected. |
correct | Which spike categories corrected replaces: recommended (default: meme,incident,investor,marketing), none, all (adds demand_event), or a comma list of meme, incident, investor, marketing, demand_event. Anything else is 400. See corrections. |
window | 7 or 28: adds window_sum (the trailing sum ending on date), base_sum (the same window 364 days earlier) and low_volume (either side under 2 posts a day: 14 or 56) to every row. Null when any day of a window is null. |
kdate … | Values as known on a past day; kdate=all for every vintage. See point-in-time reads. |
cursor | Paging. |
Entity rows: entity_id, ticker, division, date, metric, raw, corrected, flag, coverage_status, review_pending. Company-line rows (rollup=ticker) have flagged_entities instead of entity_id/division, and flag is the union of the entities' flags. Both carry company_id, source_ticker, the knowledge fields (kdate_from, kdate_to, change_reason) and ticker_at_date; columns added later are appended at the end, so positional CSV readers keep working.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/company/CELH/metrics?metric=mentions&rollup=ticker&start=2026-04-06&end=2026-04-30"
{ "as_of": "…", "embargo_cutoff": "2026-04-09", "tier": "trial", "dataset": "company_demand",
"rows": 4, "next_cursor": null, "release_id": "…", "history_mode": "…",
"requested_end": "2026-04-30", "truncated_to": "2026-04-09",
"data": [
{ "company_id": "5735", "ticker": "CELH", "date": "2026-04-06", "metric": "mentions",
"raw": 153, "corrected": 153, "flagged_entities": 0, "coverage_status": "complete",
"review_pending": false, …, "flag": "", "ticker_at_date": "CELH" },
…
{ "company_id": "5735", "ticker": "CELH", "date": "2026-04-09", "metric": "mentions",
"raw": 143, "corrected": 143, … }
] }
Trimmed; values from the published trial sample. A trial key with a 2026-04-09 cutoff asked through 04-30, so the response stops at the cutoff and says so with truncated_to (header x-truncated-to on CSV). The same happens to any key when end is past the last day the data reaches.
coverage_status is complete, missing, incomplete (no valid value that day, including a disclosed query-semantics hold), pending_review or unresolved_correction (raw present, corrected null). A company line is incomplete when any of its entities is. review_pending marks a spike that is still waiting for its editorial category; until then corrected equals raw.
GET /v1/metrics/cross-section
One metric for every entity on one date: the trailing 7 or 28-day sum of the default-corrected series. Built for screens and year-over-year ranking. /v1/demand/cross-section is the same route.
| Parameter | Notes |
|---|---|
metric | Required. One of the 16 metrics. |
window | 7 (default) or 28 days. |
date | The window's last day. Omit it for the newest stored date. A date after the newest window serves the newest one, with truncated_to and requested_date. Trial dates are clamped to the cutoff. |
core_only | true (default) or false. |
kdate … | The screen as it was published then. |
Rows: entity_id, ticker (today's), date, value (integer window total), company_id, base_date (date minus 364 days), base_sum, low_volume, ticker_at_date.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/metrics/cross-section?metric=pi&window=28&format=csv" -o pi_28d.csv
A window that contains an incomplete day returns value: null, never a partial sum. An entity's row is left out unless its whole window lies inside the entity's ownership window, and base_sum is null unless the base window does too. correct is not accepted here. Join on company_id: ticker is today's ticker whatever the date.
Consumer trends
Consumer behaviours ("cutting back on spending", "taking a trip"), each a phrase rule counting people describing doing it, daily since 2016-09-01; /v1/trends is the catalog. Trend ids (T001…) are permanent: a trend keeps its id through renames (responses always carry the current name), and a retired id is never reused. The same ids are used by the Google search trends.
GET /v1/trends
The catalog. Exactly trend_id, name, group, mapped_tickers (how many tickers the trend maps to) and mapped_datasets (how many rows it has in /v1/trends/datasets).
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" "https://data.likefolio.com/v1/trends?format=csv"
trend_id,name,group,mapped_tickers,mapped_datasets
T011,buying crypto,"Investing, Crypto & Gambling",15,0
GET /v1/trends/map
Which tickers each trend bears on. Exactly trend_id, name, ticker, weight, added_at, removed_at. weight is signed: + for a long link, − for a short one, in bands of 1.0, 0.6, 0.3 and 0.1. Every tier gets the full map.
| Parameter | Notes |
|---|---|
ticker, trend_id | Filters. |
as_of | Only links whose added_at..removed_at window contains this date (blank bounds always match), for backtesting the map as it stood. |
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" "https://data.likefolio.com/v1/trends/map?trend_id=T011&format=csv"
trend_id,name,ticker,weight,added_at,removed_at
T011,buying crypto,BTCUSD,1.0,,
T011,buying crypto,ETHUSD,1.0,,
T011,buying crypto,COIN,1.0,,
GET /v1/trends/datasets
Public data series each trend is mapped to. Exactly trend_id, name, series, sign (+1 or −1), primary, freq, desc. Filter with trend_id.
trend_id,name,series,sign,primary,freq,desc
T002,bad allergies,rs_health_personal,1,True,M,Retail sales: health and personal care stores
GET /v1/trends/{trend_id}/series
One trend's daily series. Exactly trend_id, name, date, raw, corrected, flag. Parameters: start, end, cursor. Trial ranges stop at the cutoff.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/trends/T011/series?start=2026-04-08&format=csv"
trend_id,name,date,raw,corrected,flag
T011,buying crypto,2026-04-08,294,294,
T011,buying crypto,2026-04-09,299,299,
Values from the published trial sample.
| Day | What you get |
|---|---|
| Normal day | raw and corrected are equal, flag empty. |
| Spike still under review | raw as measured, corrected null. |
| Reviewed spike kept as real demand | Both values published, flag demand_event where it applies. |
| Reviewed spike removed as noise | corrected null; flag meme when it was a meme. |
| Day declared incomplete | raw and corrected null on every trend. /v1/platform/breaks lists these days with scope trend_series. |
GET /v1/trends/{trend_id}/scores subscription
The trend's published scorecard (JSON only, no envelope). History is reconstructed and the 2024–2026 scores are fitted; correlations are not forecasts. Trial keys get 403; a trend without scores is 404.
| Field | Meaning |
|---|---|
trend_id, as_of, history_mode, fit_caveat | The trend, its series date, how its history was recorded, and the caveat on the fitted years. |
scores.days, first, last | Days in the scored series and its first and last day. |
scores.mean_daily, median_daily | Mean and median daily count of the scored (corrected) series. |
scores.mean, scores.median | Correlations of the quarterly series (mean: quarter mean of month means; median: of month medians), nested by basis (adj seasonally adjusted, raw), then window (full the whole history, test 2024–2026 fitted, ftrain 2016–2023), then fit: company (with the mapped companies' revenue growth), data (with the mapped datasets), best (the larger). A fit is null when there is nothing to correlate with. |
Data history
GET /v1/platform/breaks
Dated discontinuities in the X data, by kind, and whether a publication restated each one.
| Field | Meaning |
|---|---|
break_id, kind | kind: source_outage, source_duplication, collection_hole, partial_day or query_seam. |
date_from, date_to, utc_from, utc_to | The affected days and hours. |
metrics, scope, tickers | What is affected. tickers is a comma list when scope is tickers. |
status, restated_in | restated (a publication changed the served rows; which one) or open (served as published). |
served_as, summary | How the days are served now, and what happened. |
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" "https://data.likefolio.com/v1/platform/breaks?ticker=CAKE"
ticker keeps the breaks that affect every series plus those naming that ticker. The envelope adds breaks_updated_at.
GET /v1/changelog
Every publication and schema change, newest first (at most 500). Each company publication also says what it changed against the one it replaced: rows restated, withdrawn, removed, filled or appended, and which companies, metrics and dates.
| Parameter | Notes |
|---|---|
since | Inclusive published_at lower bound; an ISO timestamp. |
detail | release (default): one row per publication. company: one row per publication × company × metric with its date ranges. Implied by ticker, company_id or metric. |
ticker, company_id, metric | Filters. ticker matches today's ticker or the source ticker. |
Release rows: published_at, dataset, version, kind, rows_affected, summary, then release_id, change_reason, previous_release_id, parts, detail_status (computed, baseline or not_computed), rows_restated, rows_withdrawn, rows_removed, rows_filled, rows_appended, rows_annotation, companies_restated, companies_changed, tickers_restated, metrics_restated, date_from, date_to, detail_summary.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/changelog?ticker=PYPL&since=2026-10-01T00:00:00Z"
GET /v1/publications
The recorded, immutable publications the time-travel parameters select from: release_id, published_at, change_reason, manifest_sha256. Pass a release_id from here to pin a read to one publication.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" "https://data.likefolio.com/v1/publications"
Metric-specific query-semantics holds
An approved query-semantics hold withholds invalid measurements; it is not a measured zero or a source outage. In the affected CVS publication, pi, npi, ps, ns, promo and political have raw and corrected null in JSON (blank in CSV), with coverage_status=incomplete. Mentions and switching metrics are preserved. Entity/ticker rollups and 7/28-day windows must not turn held components into zero.
A company response whose selected publication carries a hold adds metric_coverage_issues, an array of objects with id, metrics (string array) and reason, plus complete_tail_scope="unheld_metrics". This is company/publication coverage metadata, so it may still appear when the request selects an unheld metric. The headers x-metric-coverage-issue: query_semantics_hold, x-held-metrics (comma-separated metric names) and x-complete-tail-scope: unheld_metrics carry the issue for CSV and other responses. Parquet/Arrow envelope metadata carries the same disclosures.
Coverage rows add metric_coverage_issue (singular; query_semantics_hold or null), held_metrics (comma-separated string or null) and complete_tail_scope (unheld_metrics or all_metrics) when the selected coverage publication contains a hold. These fields also qualify trial coverage; they do not expose brand names. A complete tail scoped to unheld metrics must not be read as completeness of all behavior metrics.
Old vintages stay old. Selecting a recorded publication from before the hold preserves its earlier values and does not retrospectively add this hold annotation. The new publication discloses the restatement; it does not rewrite old release bytes. Lifting the hold requires a separately verified replacement release, not deleting a flag or treating null as zero.