Search interest API Google search · brands and trends
Daily Google search interest for each company's brands and products, from 2020-09-01, on one common scale. You get each verified brand's series and one company line, the sum of those brands, with the company's coverage disclosed on every response except a reconstructed kdate read. The catalogs differ: 141 X trends and 139 currently served Google search trends. For Google search availability, see search trends.
History works differently by dataset. X company metrics keep dated versions (kdate) from launch; earlier history is reconstructed. Google search company series, company history and search trends keep recorded snapshots and take kdate from 2026-11-04 Eastern onward; earlier history is reconstructed. /v1/search/brands, /v1/search/trends/map and bulk ZIPs serve the latest only and reject kdate. Ownership windows govern which brands count toward a company. See Google search at a kdate.
404 not covered. The rule, and how to read coverage, are in verified brands and coverage.Company and brand series
GET /v1/company/{ticker}/search
One row per verified brand per day, or with rollup=ticker one company line per day.
| Parameter | Notes |
|---|---|
ticker (path) | The company's ticker, e.g. NKE. Case-insensitive. |
rollup | brand (default) or ticker. |
start, end | Inclusive days. No row is ever later than the company's complete_through; trial ranges also stop at the embargo cutoff. |
format, cursor | See formats and paging. |
kdate, history_mode | Read a recorded or reconstructed snapshot. See point-in-time reads. |
Brand rows (rollup=brand)
| Field | Meaning |
|---|---|
brand | The brand or search term. Trial keys get brand_id (TICKER#bN) instead, never the name. |
ticker, date | The company and the day. |
interest | Search interest on LikeFolio's common scale (the smallest brand averages about 50; no ceiling), not Google's 0–100. A string with 2 decimals on brand rows. |
valid_from, valid_to | The brand's ownership window; blank means open. No row falls outside it. See ownership windows. |
resolution | daily (built from Google's daily values) or weekly (built from Google's weekly values; compare weeks, not days). See resolution labels. |
provenance | A per-day record of how the value was produced. Not needed to use the data. |
Company-line rows (rollup=ticker)
ticker, date, interest (the sum of the verified brands that day, a number) and brands (how many brands had a value that day). A day is in the line only through complete_through, but the current cutoff does not guarantee that every expected brand has refreshed.
Envelope fields
| Field | CSV header | Meaning |
|---|---|---|
complete_through | x-complete-through | The company's pipeline-estimated cutoff day. No row is later. |
coverage | x-search-coverage | The share of the company's search interest over the trailing 365 days (to complete_through) that comes from its verified brands, 0–1. Null if the company had no search interest in that window. |
coverage_full_history | The same share since 2020-09-01. | |
verified_brands | x-search-verified-brands | How many brands you receive. |
candidate_brands | How many brands we track for the company, verified or not. |
# every verified brand, daily
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/company/NKE/search?start=2025-01-01&format=csv" -o nke_brands.csv
# the company line
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/company/NKE/search?rollup=ticker&start=2025-01-01"
{ "as_of": "2026-10-31", "embargo_cutoff": null, "tier": "subscription", "dataset": "search_interest",
"rows": 669, "next_cursor": null,
"complete_through": "2026-10-31", "coverage": 0.3423, "coverage_full_history": 0.3991,
"verified_brands": 7, "candidate_brands": 10,
"data": [
…
{ "ticker": "NKE", "date": "2026-10-30", "interest": 25459.11, "brands": 5 },
{ "ticker": "NKE", "date": "2026-10-31", "interest": 26843.42, "brands": 5 }
] }
// rollup=brand, subscription key (one row)
{ "brand": "Air Jordans", "ticker": "NKE", "date": "2020-09-01", "interest": "19513.72",
"valid_from": "", "valid_to": "", "resolution": "daily", "provenance": "…" }
// the same row for a trial key
{ "brand_id": "NKE#b7", "ticker": "NKE", "date": "2020-09-01", "interest": "19513.72", … }
Illustrative values from a pre-launch rehearsal; live values differ. In this example 7 of Nike's 10 tracked brands are verified and carry 34% of its trailing-year search interest, so the company line represents about a third of Nike's tracked search demand.
A company with no verified brand, below the 25% floor, or not tracked at all: 404 {"error": "not covered: TICKER"}.
Brands and shares
GET /v1/search/brands
The verified brands behind each company's line: one row per brand, with its share of the company's verified search interest over a window and the company's coverage. Latest snapshot only: kdate answers 400.
| Parameter | Notes |
|---|---|
ticker | Required. One ticker or a comma list. |
start, end | The share window. Default: the 365 days ending at the company's complete_through. Trial windows stop at the embargo cutoff. |
format | As usual. |
| Field | Meaning |
|---|---|
ticker, brand, brand_id | Trial keys get brand_id only. Subscription keys get both, which is how a trial model joins to brand names. |
valid_from, valid_to | The brand's ownership window. |
first_date (first non-null value on or before the last complete day; explicit zero counts), last_date | The brand's first and last served day. |
window_start, window_end, window_days | The share window and how many days in it had a value. |
complete_through | The company's pipeline-estimated cutoff day. |
mean_interest, share | The brand's mean interest in the window, and its share of the company's verified interest. Shares add up to 1 per ticker. |
days_outside_window | Days in the share window outside the brand's ownership window (not counted). |
verified, coverage | Always true here; the company's coverage. |
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/search/brands?ticker=NKE,AAL"
{ "dataset": "search_brands", "rows": 13, …,
"data": [
{ "ticker": "NKE", "brand": "Air Jordans", "brand_id": "NKE#b7", "valid_from": "", "valid_to": "",
"first_date": "2020-09-01", "last_date": "2026-10-31", "window_start": "2025-11-01",
"window_end": "2026-10-31", "complete_through": "2026-10-31", "window_days": 365,
"mean_interest": …, "share": …, "days_outside_window": 0, "verified": true, "coverage": 0.3423 },
…
] }
Trimmed and illustrative. share × coverage is roughly the brand's share of all the company's tracked search interest.
Search trends
The consumer behaviours of the X trends, measured in Google search instead. Each trend is the sum of 2 to 8 Google search terms, and each trend was reviewed against Google before it ships. A trend is published only when it matches Google and none of its terms is an unpublished brand, the wrong entity, or missing; any other trend is not served.
- Shared IDs, source-specific coverage. Where a Google search trend and an X trend share a
trend_id, they describe the same behaviour. Each source has its own catalog; a trend may be available in only one source. Join only IDs present in both catalogs, rather than assuming every Google search trend has an X counterpart. - Google's values as Google shows them, daily, with no correction.
- Each current series stops at its selected endpoint (
complete_through).
GET /v1/search/trends
The catalog of published search trends. Exactly trend_id, name, first_date (first non-null value on or before the last complete day; explicit zero counts), last_date (the trend's last complete day), members (how many search terms it sums) and resolution (daily). Parameters: format, kdate, history_mode (see point-in-time reads).
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" "https://data.likefolio.com/v1/search/trends?format=csv"
GET /v1/search/trends/map
Which companies each search trend bears on: the X trends map (same columns, same signed weights) limited to the published search trends. Filters: ticker, trend_id. Latest only: kdate answers 400, because the map comes from the X data.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" "https://data.likefolio.com/v1/search/trends/map?trend_id=T007"
GET /v1/search/trends/{trend_id}/series
One search trend's daily series. Exactly trend_id, name, date, interest, one row per day through the trend's last complete day. The envelope carries complete_through (CSV header x-complete-through) and resolution. Parameters: start, end, cursor, format, kdate, history_mode (see point-in-time reads). Trial keys get no day after the embargo cutoff. A trend that is not published is 404.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/search/trends/T007/series?start=2026-10-01"
{ "dataset": "search_trends", "complete_through": "2026-10-30", "resolution": "daily", …,
"data": [ …, { "trend_id": "T007", "name": "buying a home", "date": "2026-10-30", "interest": 41.0 } ] }
Trimmed; the value is illustrative. /v1/status lists search_trends with its as_of. Without kdate, search trends serve the latest publication. They have no bulk drop.
Point-in-time reads
Four routes take kdate and history_mode: /v1/company/{ticker}/search, /v1/company/{ticker}/search/history, /v1/search/trends and /v1/search/trends/{trend_id}/series. They are recorded from the first snapshot, which kdate 2026-11-04 Eastern reads; earlier history is reconstructed from that snapshot and labelled. Not supported at launch: /v1/search/brands, /v1/search/trends/map, bulk ZIPs, observed_at and kdate=all.
- 27 held companies. Those held in the first snapshot answer
404on/v1/company/{ticker}/searchat everykdateread from a snapshot that holds them, including dates when they were live. They are reachable only through/v1/company/{ticker}/search/history. - The row dated
kdate. A reconstructed read includes it, a look-ahead of at least one day. X reconstructions use the same rule. - Trial keys. The embargo cutoff still applies at every
kdate.
Selection, labels, error answers and the other limits are in Google search at a kdate.
curl -s -H "Authorization: Bearer $LIKEFOLIO_API_KEY" \
"https://data.likefolio.com/v1/search/trends/T007/series?kdate=2026-11-04&start=2026-10-01"
{ "dataset": "search_trends", "ownership_kdate": "2026-11-04", "history_mode": "recorded",
"release_id": "gt-vintage-2026-11-05T04:20:42Z", "kdate_from": "2026-11-05T04:20:42Z",
"recorded_from": "2026-11-05T04:20:42Z", …, "data": [ … ] }
Trimmed. kdate_to is the next snapshot's capture time, or null.
Everything at once
The search bulk drop (brands, not trends) holds one CSV per published company plus search_coverage.csv (one row per company: ticker, coverage, coverage_full_history, verified_brands, candidate_brands). List it with GET /v1/bulk?dataset=search; see bulk downloads.
Observed, filled and transformed values. Some values are filled or rescaled rather than read straight from Google. The provenance and resolution columns tell you which: lf_fill marks a value we filled in, and weekly values may be spread across daily rows. A missing row means no data, not zero.
Display names are not query specifications. A public brand name can remain unchanged after its underlying text query is replaced by a Google Topic. The name alone does not establish which Google query or Topic produced the series.
Reading the numbers
- Compare a brand with itself over time. Levels across brands are approximate.
- Unavailable output days are omitted. Existing pipeline-filled values remain visible and are identified by provenance; missing output is not evidence of a Google zero.
- No current series runs past
complete_through, and brands count only while the company owns them. - Lower coverage means a narrower line. The company line is built from verified brands only. Check
coveragebefore comparing companies. - Coverage changes as brands are verified. A newly verified brand is added with its full history, and the company's coverage moves.
Event time is not availability time. Weekly values show when people searched, not when we received the data. Before the first recorded snapshot (kdate 2026-11-04 Eastern), search history is reconstructed from the latest data rather than a true point-in-time record; from that date on, each day's snapshot is kept.
Company identity on search routes
Client Google search routes use company-lifetime aliases for two explicitly mapped companies. VSCO and VSXY select company 27124; output uses canonical ticker VSXY, with eligible company history beginning July 21, 2021. BBBY selects current company 28701 only, using its approved brand ownership windows; it does not combine history from unrelated companies that reused that symbol.
For these two mappings, JSON responses from /v1/company/{ticker}/search add envelope fields company_id (string), canonical_ticker and search_alias_policy (company_lifetime). These are optional fields, not new brand-row columns or CSV headers. Other companies retain their existing response shape. Trial brand IDs use the canonical ticker namespace; the aliases do not create a second set of brand IDs.
/v1/search/brands accepts the same mapped ticker aliases and deduplicates aliases selecting the same company; its envelope does not add these three fields. Company IDs and private storage identifiers are not new public search aliases. Internal raw search keeps its existing behavior and gains no newly supported aliases. X company identity, dated ticker resolution and /v1/resolve are unchanged. Company-lifetime routing does not make Google history point-in-time; kdate does that from 2026-11-04 Eastern onward.
Source snapshot date is not completeness. The search_interest entry in /v1/status.datasets reports as_of as the source snapshot date, which may be intraday. It is not a global complete day. Status does not provide a Google-search complete_through field. Use each current company response’s complete_through for its closed-UTC, active-basket endpoint; the history route returns complete_through=null.
Preserved company search history
GET /v1/company/{ticker}/search/history explicitly reads the preserved legacy snapshot, including for a company whose current series is held. It uses the same authentication, entitlement, client company aliases, identity/ownership windows, trial embargo and brand-row projection as company search. Parameters are start, end, rollup=brand|ticker, cursor, format, kdate and history_mode; invalid dates, reversed ranges, rollups, cursors or formats return 400. Unauthenticated requests return 401. Private storage identifiers remain unavailable.
Without kdate, the JSON envelope reports complete_through: null, current_freshness: "not asserted", basket_completeness: "not certified; historical missing-member days retained", history_mode: "legacy reconstructed snapshot; no vintage/PIT claim" and weekly_shape: "legacy interpolated/carried values may be present". These are response disclosures. With kdate, the envelope keeps the other four and history_mode is recorded or reconstructed; recorded says which bytes were read, not that the basket is certified. Historical rollups may contain smaller baskets. Preserving the archive does not make every raw row visible: normal authorization and identity/tier projection still apply. This is not a native-weekly serving route.
Held current series
If no complete active basket exists within that bounded window, the current route returns HTTP 404 with error="search_series_held", data=[], complete_through=null, ticker, closed_calendar_day, reason="no_complete_active_basket_within_freshness_window", missing_members (omitted for trial keys), historical_snapshot_available=true and historical_route. The hold applies to that company only. Current requests do not silently fall back to historical CSVs. Use the explicit history route when its disclosure is acceptable. At a kdate, see held companies.
Ready current JSON responses without kdate add completeness_scope, closed_calendar_day and history_mode. Existing coverage remains the trailing-interest measure, not certification of every historical basket. A company's current complete_through remains subject to identity and tier restrictions.
Inactive-current hold. An approved registry-ended company uses reason=company_registry_ended and company_registry_end (the approved date), with complete_through=null and the explicit historical_route. Its response omits missing_members for every tier. This differs from a recent-basket missing-data hold. A READY source endpoint later than the snapshot’s recorded closed_calendar_day or more than 14 days before that recorded day is malformed availability metadata and fails with 503; it is not silently relabelled as an ordinary company hold.
Display labels and identity. Subscription company-search and explicit history JSON envelopes may include brand_display_names, keyed by stable brand ID. Each entry contains canonical_brand and friendly display_name. Trial responses omit this metadata. Canonical CSV labels, column layouts, brand/brand_id values and stable IDs remain unchanged. Friendly names are presentation metadata only; they do not change the query, numeric history, ownership or eligibility.
Inactive companies and containment. Approved inactive-current holds return a company-local current 404 while preserving explicit verified history access. TGT containment removes the unresolved generic target row from the current CSV/ZIP projection only, preserving its original rows in explicit history under normal identity/ownership/tier restrictions and reconstructed/uncertified disclosures. It does not substitute a new Target query into old values. TGT current availability still requires the recomputed coverage and active basket to pass. A trial embargo alone is not a reason to hold a current company.
Pagination and publication consistency
Treat next_cursor and x-next-cursor as opaque strings. Continue with the same endpoint, parameters, format and key tier. Company-search cursors bind to the current publication (at a kdate, the selected snapshot) and embargo cutoff; after a publication, query, format or tier change, the API returns HTTP 400 with field=cursor. Discard the unfinished pagination result and restart without a cursor. This applies to current and explicit-history company search; a cursor from one route cannot continue the other.
During the short GT publication swap, a request may receive HTTP 503; retry the whole request with backoff. The service does not substitute historical or partially published data. If a subsequent cursor request returns 400, restart pagination without the old cursor. This transport-consistency guard does not add a live-clock freshness rule: closed_calendar_day remains the recorded snapshot clock. It does not certify source completeness or final launch readiness.