Prise en main

Quickstart

Base URLs

Use an environment variable so scripts can switch base URLs without changing code:

export TT_API_BASE_URL="https://api.trendtrack.io"

30-second sanity check

Run this in a terminal — it authenticates, resolves your workspace, and returns real JSON in under a second. No credits are spent.

export TT_API_KEY="tt_live_…"
export TT_API_BASE_URL="https://api.trendtrack.io"
curl -s "$TT_API_BASE_URL/v1/workspace" \
  -H "Authorization: Bearer $TT_API_KEY" \
  -D - | head -20

{
  "data": {
    "id": "ws_01HZX0W3YJABCDEF",
    "name": "Acme Growth",
    "plan": "professional",
    "createdAt": "2025-09-12T10:22:31.000Z"
  },
  "requestId": "req_01HZ…"
}

Relevant response headers returned by the curl -D - flag:

• X-Request-Id — quote this when reaching out to support.
• X-Credits-Remaining — sent by metered endpoints and by unmetered GET /v1/usage, which reports the current balance.

Shell endpoints (/v1, /v1/health, /v1/me, /v1/workspace, /v1/usage) are unmetered. They are the safest way to validate authentication, request IDs, and workspace resolution before moving on to metered business endpoints. GET /v1/usage still returns X-Usage-Cost: 0, X-Credits-Used: 0, and X-Credits-Remaining.

A successful response always carries { data, requestId } (plus pagination on collection endpoints). Errors follow the same shape — always parse error.code, not the human message:

{
  "error": {
    "code": "missing_api_key",
    "message": "Missing API key.",
    "requestId": "req_123"
  }
}

Common first errors

• 401 missing_api_key / 401 invalid_api_key: header missing or malformed. Double-check Authorization: Bearer <key> (no x-api-key, no quotes around the value).
• 401 credential_revoked / 401 credential_expired: the key was deleted or rotated — issue a new one in workspace settings.
• 403 workspace_public_api_disabled: Public API access has not been explicitly enabled for the workspace yet. A missing access row is treated as disabled.
• 402 insufficient_credits: call GET /v1/usage to inspect the remaining balance, then top up or wait for the monthly refill.
• 429 rate_limited: respect the Retry-After header before retrying. See Rate Limits.

Full list and status-code guidance lives in Errors.

Pagination pattern

Collection endpoints accept limit and offset (or a page parameter — check the operation page). Cost is charged per returned row, not per request, so always page conservatively:

curl -X GET "$TT_API_BASE_URL/v1/ads?search=skincare&limit=50&offset=0" \
  -H "Authorization: Bearer $TT_API_KEY"

GET /v1/ads always requires search. If you need advanced filters with no text term, use POST /v1/ads/query with a JSON body instead.

• Start with a small limit (10–50) while integrating to avoid burning credits on a loop bug.
• Watch the X-Credits-Remaining response header after every metered call.
• Stop paginating when the server returns fewer rows than limit.

Track usage as you go

Metered responses can include:

• X-Usage-Cost — credits consumed by that request
• X-Credits-Used — credits consumed by that request
• X-Credits-Remaining — balance after the call
• X-Credits-Source — recurring, topup, or hybrid when a bucket is consumed
• X-Request-Id — include it when contacting support

GET /v1/usage is unmetered, but it still returns X-Usage-Cost: 0, X-Credits-Used: 0, and X-Credits-Remaining.

See Credits & Billing for the complete consumption model.

Agent integration path

For tracked-brand questions, resolve the workspace brandtracker first instead of searching the public ad library:

1. GET /v1/me — validate the key and workspace.
2. GET /v1/usage — check credit budget before metered loops.
3. GET /v1/system/freshness (optional) — read latestReadyDate and dataFreshnessLagDays before reporting historical, freshness-sensitive metrics.
4. GET /v1/brandtrackers?page=1&limit=100 — match the user's brand name against data[].name and carry data[].id as brandtrackerId.
5. GET /v1/brandtrackers/{brandtrackerId}/overview — start with the overview for a broad read, then use canonical top-ads ranking calls when the user asks for winners or movers.
6. GET /v1/brandtrackers/{brandtrackerId}/top-ads?sortBy=currentRank — current ranking. Use sortBy=rankDelta7d, rankDelta14d, or rankDelta30d for rank movers, and sortBy=reach, reachDelta1d, reachDelta7d, or reachDelta30d for reach rankings.

Do not add period or snapshotDate to canonical top-ads ranking calls; they are deprecated compatibility parameters on /top-ads. /scaling-ads and /ad-rank are legacy compatibility endpoints for rank movers and may return an empty trajectory on the ES-backed path.

Key concepts

A few field names worth knowing before browsing the reference:

• rank / currentRank — the current Facebook page rank returned on top-ads responses. Sort canonical ranking calls with sortBy=currentRank.
• rankDelta7d / rankDelta14d / rankDelta30d — canonical top-ads sort keys for rank movers. Use these instead of starting new integrations on /scaling-ads or /ad-rank.
• reach — the Meta Ad Library term for the public-facing impressions-style metric, exposed directly by our API. On brandtracker/top-ads GET endpoints, sort with sortBy=reach, reachDelta1d, reachDelta7d, or reachDelta30d; impressions is accepted as a reach alias. On POST /v1/ads/query, use sortBy: "reach" with order.
• daysRunning — number of calendar days an ad has been observed live. Useful as a longevity signal (winners typically stay live > 14 days).
• firstSeenAt — first timestamp at which Trendtrack observed the ad. Combined with daysRunning, it lets you reconstruct the live window.
• snapshotDate — an anchor date for historical/freshness-aware endpoints. It is deprecated and ignored on canonical /top-ads ranking calls. See Time windows for where period and snapshotDate still apply.

Next steps