Troubleshoot missing data

When a chart is empty, debug from the ingestion response toward the visualization. Changing multiple filters at once usually hides the actual cause.

1. Check the HTTP response#

Log the response status and body from the ingestion request:

const response = await fetch("https://api.graphjson.com/api/log", options);
const body = await response.text();

console.log(response.status, body);

A successful network call is not the same as a successful ingestion. Treat only a 2xx status as accepted.

Common failures:

2. Confirm the collection#

The request collection must exactly match the collection you are viewing:

{
  "collection": "product_events"
}

Collection names are lowercase in the dashboard. Check for singular/plural mismatches and whitespace in code or environment configuration.

If you are unsure, inspect all_events to determine whether the event reached the workspace under another collection.

3. Open Samples#

Use:

• visualization: Samples
• range: 7 days ago → now
• order: Descending
• filters: none

Samples removes aggregation and split logic from the problem. If the event appears, ingestion worked and the issue is in the chart configuration.

4. Check event time#

A request can succeed while its event falls outside the visible range.

Confirm the timestamp is Unix seconds:

console.log(new Date(timestamp * 1000).toISOString());

For webhooks and backfills, confirm you are using the source event time you intended. Check whether the dashboard’s IANA time zone moves the event across a calendar boundary.

5. Check field names and types#

Field names are case-sensitive. These are different:

user_id
userId
User_ID

Numeric aggregations require JSON numbers:

{"amount": 4900}

not:

{"amount": "4900"}

Nested fields are flattened by default, so context.page.path is the field name shown by the visualizer.

6. Reduce the visualization#

Rebuild the chart in this order:

1. Count
2. no metric
3. no split
4. no comparison
5. no filters
6. broad range

Then add one setting at a time. The first change that empties the result identifies the problem.

7. Inspect the generated API call#

Use Export → As Data API Call and compare the generated payload with the chart controls. Look for:

• an old collection name
• a stale or overly narrow time range
• filters with the wrong type
• a metric that is not numeric
• a split with no matching values
• an unintended compare

8. Reproduce with SQL#

Run:

SELECT
  timestamp,
  json
FROM product_events
ORDER BY timestamp DESC
LIMIT 50

Then add the failing condition directly:

SELECT count()
FROM product_events
WHERE JSONExtractString(json, 'event') = 'checkout_completed'

SQL separates the stored data from visualizer configuration and makes type mismatches visible.

Contact support#

If the event has a 2xx ingestion response but cannot be found, email hi@graphjson.com with:

• approximate request time and time zone
• endpoint
• collection name
• HTTP status and response body
• a redacted example payload

Do not include the API key or sensitive event values.