guides / kick-stream-analytics
Kick analytics are a sample, not a census.
Kick does not provide a cursor that enumerates every livestream. TrackTron therefore combines several bounded query strategies and publishes the resulting coverage limits beside the data.
How TrackTron discovers Kick streams
Each Kick livestream query returns at most 100 streams. TrackTron combines global queries, category queries, language queries, and refreshes of previously known channels. It varies sort order and records whether a query hit the result cap.
The normal plan runs on an approximately ten-minute cadence, queries a rotating set of categories and languages, and refreshes known channels. Less recently successful query targets receive priority rather than repeatedly walking the same mutable result list.
For each discovered stream, TrackTron preserves which query sources found it. That provenance allows the coverage dashboard to report query runs, capped queries, errors, and the ratio between observed streams and Kick's reported live count.
Reported live is not observed viewers
Kick Reported Live comes from Kick's livestream statistics endpoint and represents Kick's reported stream count. Observed streams are the unique streams returned by TrackTron's query plan. Observed viewers are the viewers attached to that observed subset.
Kick does not provide a matching platform-total viewer count. Do not present observed viewers as all viewers on Kick, and do not divide observed viewers by reported live streams as if both values describe the same population.
When a query reaches its 100-stream cap, relevant streams may exist beyond the returned page. Category, language, and known-channel queries recover additional coverage, but they do not prove completeness.
What channel and category metrics mean
Tracked Sample Average
Kick channel averages are observation-weighted over the samples TrackTron retained. They are not official Kick creator analytics and are not duration-weighted continuous measurements.
Tracked peak
The peak is the highest observed value in retained samples. A higher value may have occurred between observations or during a period when the channel was not rediscovered.
Tracked rankings
Rankings compare channels inside the tracked dataset. They are not platform-wide Kick rankings. The observed ecosystem is biased toward streams likely to appear in global and high-activity query results.
Session timing
A session end is a last-observed bound rather than an exact offline timestamp. Discovery gaps are missing data, never zero-viewer measurements.
Where the sample is strongest and weakest
Viewer-count sorting makes high-viewer streams easier to observe. Category and language queries broaden the sample, while known-channel refreshes improve continuity for channels already encountered. Newly created, low-viewer, or infrequently queried streams are easier to miss.
A live row can remain in the rotating table beyond the freshest live threshold, so the interface distinguishes fresh live state from recent discovery history. Sparse channel pages explicitly show observation counts and date coverage.
Historical Kick data is also downsampled. Old platform observations become daily points and old per-stream observations become hourly points. Long-range charts therefore have less resolution than recent windows.
How to compare safely
- Compare Kick channels only when both have meaningful sample counts and similar date coverage.
- Use broad directional patterns rather than precise market-share claims.
- Read the live coverage panel before interpreting ecosystem concentration.
- Do not compare Twitch's cursor-based enumeration directly with Kick's bounded observed subset.
- Do not convert missing observations into zeros or exact offline times.
The live Kick Coverage Methodology page publishes current query statistics alongside the contract described here.