Analytics

Toggld counts flag evaluations per flag, environment, variant, and reason. Use the counts to confirm that your application reads a flag, to check a rollout's variant mix, and to find errors.

View analyticsCopy Anchor Link

On a flag page. The Evaluations panel shows the last 7 days for the selected environment:

• The total number of evaluations.
• The share of evaluations that served each variant.
• The number of evaluations that returned an error, when there are any.

On the Analytics page. Open Analytics in your workspace, then choose a flag, an Environment, and a Range: Last hour, Last 24 hours, or Last 7 days. The page shows evaluations over time and the served variants.

Every role, including readers, can view analytics.

Where the counts come fromCopy Anchor Link

What Toggld storesCopy Anchor Link

Toggld stores grouped counts only. Each count covers one flag, environment, variant, reason, and UTC minute.

Toggld does not store user identities, evaluation contexts, flag values, or metadata in analytics.

By default, Toggld keeps counts for 30 days.

Timing and accuracyCopy Anchor Link

Counts are approximate and can lag behind your traffic:

• The API records counts in the background, after it sends the evaluation response.
• Local SDKs send counts every 30 seconds, so the dashboard can be up to 30 seconds behind, plus processing time.
• During an outage, or when a process crashes, buffered counts can be lost.
• When a buffer is full, new groups are dropped.

If a flag shows no evaluations, check that local SDKs have telemetry turned on and that they flush before they exit. Short scripts must call close() or flushTelemetry().

Read analytics with the APICopy Anchor Link

GET /api/v1/flags/{key}/analytics returns the data for one flag in the key's environment. It requires a management key. The range parameter accepts 1h, 24h, or 7d. The default is 24h.

curl "https://api.toggld.io/api/v1/flags/new-checkout/analytics?range=24h" \
  -H "Authorization: Bearer ffk_YOUR_MANAGEMENT_KEY"

{
  "range": "24h",
  "from": "2026-10-30T12:00:00.000Z",
  "to": "2026-10-31T12:00:00.000Z",
  "total": 1520,
  "errors": 3,
  "lastEvaluatedAt": "2026-10-31T11:58:00.000Z",
  "timeline": [{ "bucketStart": "2026-10-31T11:00:00.000Z", "count": 64 }],
  "variants": [{ "variant": "on", "count": 312 }, { "variant": "off", "count": 1205 }]
}

Analytics reads have a lower rate limit than other reads: 30 requests per minute per environment.

Send counts from a custom clientCopy Anchor Link

If you build your own local evaluator, send grouped counts to POST /v1/telemetry with a secret key or a public token. Toggld takes the workspace and environment from the credential.

curl https://api.toggld.io/v1/telemetry \
  -H "Authorization: Bearer ffk_YOUR_SECRET_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "sdkInstanceId": "checkout-api-7f3c",
    "windowId": "flush-000123",
    "rollups": [
      { "flagKey": "new-checkout", "variant": "on", "reason": "SPLIT", "bucketStart": "2026-10-31T12:00:00.000Z", "count": 42 }
    ]
  }'

The endpoint returns HTTP 204 when it stores the batch.

Rules for safe retries:

• To retry a batch, send the same sdkInstanceId, windowId, and payload. Toggld counts each instance and window pair once per environment.
• Use a new windowId for each batch of new evaluations.
• Toggld drops counts that are older than the retention period or, by default, more than five minutes in the future. It still returns HTTP 204.