Live series

curl --request GET \
  --url 'https://truestat.io/api/v1/sites/{siteId}/realtime/series' \
  --header 'Authorization: Bearer <token>'

{
  "data": [
    {
      "goals": 0,
      "views": 2,
      "minute": "2026-09-16 14:01:00",
      "bounces": 0,
      "indexing": 3,
      "sessions": 1,
      "training": 5,
      "visitors": 1,
      "answering": 0,
      "ai_referrals": 0,
      "new_visitors": 1,
      "crawler_views": 8,
      "revenue_cents": 0,
      "returning_visitors": 0,
      "avg_session_seconds": 34
    }
  ],
  "meta": {
    "minutes": 30,
    "site_id": "6fed8980-a119-4a9c-a4d4-53e448ef9f9e"
  }
}

Minute-by-minute activity for the recent past, humans and crawlers on the same rows.

Takes minutes rather than a window — see online now. Thirty by default, which is what the dashboard's live chart draws.

Quiet minutes have no row

This is the part that catches people out.

The query groups by minute, so a minute in which nothing happened is absent rather than returned as zeroes. Asking for 30 minutes on a quiet site returns however many minutes saw activity — nine, say — and they are not evenly spaced.

Plot those rows straight onto a chart and you get a lie: the quiet stretches compress to nothing and the traffic looks steadier than it was. Fill the gaps before drawing, with whatever your chart should show — a zero, a break in the line, an interpolation.

We do not fill them here on purpose. Doing so would mean returning rows the query never produced, and how a gap should be drawn is a decision for the chart rather than the transport.

There is no unit

Buckets are always one minute. A live chart that buckets by hour is not a live chart, and timeseries already buckets by anything larger.

It also needs no timezone handling, which /timeseries does: a minute starts at the same instant in every zone, so there is no boundary to get wrong.

Both sides, one timeline

The human fields exclude crawlers. crawler_views counts them and splits three ways — answering, indexing, training — for the same reason crawler series does: a fetch answering somebody now and a fetch harvesting pages for training are different events.

They share these rows because a live view draws both on one timeline.

What you can build with it

Everything the dashboard's live view shows: the visitor line, the new-versus- returning split, sessions and bounce, goals and revenue as they land, and the crawler layers underneath. It is one query per refresh rather than one per series.

Authorizations

Path parameters

Query parameters

Response