Kick Info

Glossary

What every metric on Kick Info actually means.

Last updated May 5, 2026

Different analytics dashboards use the same words to mean slightly different things. This page nails down exactly what each metric on Kick Info represents and how it’s computed, so two people looking at the same number arrive at the same understanding.

Average viewers

Time-weighted average of concurrent viewers across the chosen window. For a single stream: area under the viewer-count curve divided by stream duration. For a channel over 7 / 14 / 30 days: total viewer-hours divided by total broadcast hours, so long streams have proportionally more weight than short ones. For a language ranking: total viewer-hours from every channel in that language divided by the number of hours in the window (168 for 7 days, 720 for 30 days). All averages are weighted by time, never raw arithmetic means.

Peak viewers (this stream)

The highest concurrent-viewer count observed across the duration of a single stream. Computed as the maximum of every viewer-count snapshot taken while the stream was open. Because snapshots are taken every ~2 minutes, the recorded peak can miss a very brief spike that occurred between samples — the true peak is at most a couple of percent higher.

All-time peak

The highest peak-viewer count we’ve ever recorded for a channel across every stream they’ve broadcast since we started tracking them. Does not predate the moment we discovered the channel.

Hours watched

Total viewer-time across a stream or window. A two-hour stream with 1,000 concurrent viewers produces 2,000 hours watched. We compute this as the area under the viewer-count curve and report it directly. Hours watched is the metric sponsors care about because it captures audience size and session length together.

Hours streamed

Total broadcast time. If a channel streamed for two hours, that’s two hours streamed regardless of how many viewers they had. Sum of all session durations in the window.

Followers

Total follower count on Kick.com at the most recent moment we sampled the channel. Live channels are sampled every 5 minutes, non-live channels rotate through a slower queue (so a non-live channel’s “total followers” can lag the true count by a few hours).

Followers gained

Delta in follower count over the chosen window. For a stream it’s the difference between samples bracketing the broadcast. For a 30-day window it’s the difference between the latest sample and the sample taken 30 days ago. When no sample exists exactly at the window boundary, the nearest snapshots before and after are used; this introduces at most ~5 minutes of attribution error per stream.

Followers gained (per category)

On a channel’s “Most streamed games” table, followers gained is the session-level delta attributed to each category in proportion to how much of the stream the channel spent on it. If a streamer gained 1,000 followers across a 5-hour broadcast split as 4 hours of Just Chatting and 1 hour of GTA, Just Chatting is credited with +800 and GTA with +200.

Live now / Live channels

Channels currently broadcasting at the moment of the snapshot. Deduplicated to one row per channel (occasionally a stream-restart race produces two open sessions briefly; we collapse them). Excludes channels filtered by the bot rule below.

Bot filter

Channels whose current viewer count exceeds their total follower count are hidden from all live displays and rankings. A 6,000-viewer channel with 12 followers is almost certainly view-botted; real audiences correlate with real follower counts. This removes about 6% of nominally-live channels from public views.

Partner

Kick’s top-tier creator program: channels with the partner badge on Kick.com get a green shield on Kick Info rankings. Partner status is read from Kick’s channel endpoint and refreshed when the channel’s metadata is re-fetched (every six hours at minimum).

Verified

Kick’s identity-verification badge, applied independently of partner status. Verified channels have confirmed their identity with Kick. Less common than the partner badge.

Total views

The all-time view counter Kick.com displays on each channel profile. Reflects cumulative views since the channel was created on Kick, including views from before we started tracking. We refresh it on every follower sample.

Avg concurrent viewers (language)

Total viewer-hours from every channel tagged with that language, divided by hours in the window. So a language with many low-viewer streams and a language with few high-viewer streams can produce the same average — the metric captures total audience attention, not number of channels.

Avg concurrent channels (language)

Total broadcast-hours from every channel tagged with that language, divided by hours in the window. Tells you how many channels in that language were typically live at any given moment across the window.

Range (7D / 14D / 30D / All)

The lookback window for time-windowed metrics. “7D” means “the last 7 days”, ending now. “All” means every record we have for the channel, from the moment we discovered them to now. Range only affects time-windowed columns; all-time peak and current follower count are the same regardless of which range is selected.

Uptime

For a stream that’s currently live, the elapsed time since the stream started, displayed on channel pages. Computed as now − stream.startedAt.

Discovery delay

The gap between when a streamer goes live and when we first record a viewer snapshot for them. Usually 0–5 minutes depending on where they fall in Kick’s live directory when our walk runs. Rarely longer than 10 minutes.

Materialised table

A pre-computed table that the collector rebuilds on a cron schedule. Used for the rankings, language splits, live snapshot and home-page KPIs so pages render in milliseconds rather than running the heavy aggregation per request. See the methodology page for details.

Cache-control / Edge cache

Cloudflare’s caching layer. Each unique URL is cached independently with a TTL between 30 seconds (home) and 1 year (immutable images). The cache is what makes the site survive traffic spikes — see methodology for the per-path TTLs.