VSPO! Collaboration Map

Select a bubble and the map re-centres on that member, with closer connections drawn larger and further inside.

32 of 32 members, 22017 streams analysed (last updated 2026-09-19 06:08:18 UTC)

The same figures as the chart, in text. The total adds up the mentions of every pair that member appears in.

External VTubers and streamers mentioned on VSPO! members' streams (counted when mentioned in 2+ streams by the same member).

For each member (all 32 members across VSPO! JP and VSPO! EN: Sumire Kaga, Nazuna Kaga, Toto Kogara, Uruha Ichinose, Noah Kurumi, Mimi Tosaki, Sena Asumi, Hinano Tachibana, Lisa Hanabusa, Ren Kisaragi, Qpi Kaminari, Beni Yakumo, Ema Aizawa, Runa Shinomiya, Tsuna Nekota, Ramune Shiranami, Met Komori, Akari Yumeno, Kuromu Yano, Kokage Tsumugi, Yuuhi Sendo, Hanabi Choya, Moka Amayui, Saine Ginjo, Chise Tatsumaki, Remia Aotsuki, Arya Kuroha, Jira Jisaki, Narin Mikure, Riko Solari, Eris Suzukami, Juno Umezono), the most recent streams and videos (up to 1000) are read from their YouTube channel through the official API, and the count is how often the title or description names the other member or links their channel. The weight of a line is the sum in both directions. The number of streams analysed varies between members (26 to 1000), so take care when comparing them.

Colours come from the colour-code list on the unofficial wiki.

The picture inside each bubble is that member's YouTube channel icon. The map simply points at the image URL returned by the official API: this site neither copies nor redistributes the images — your browser loads them straight from YouTube each time. All rights in those images remain with their owners. Members whose icon has not been collected yet keep a plain circle in their image colour.

⚠ These are mention counts, not official collaboration counts. This site is an unofficial fan page with no connection to VSPO! or its operating company.

In the full map, bubble size is a member's total mention count and line thickness and opacity show the count between two members. The heaviest pair is drawn thickest and darkest; sparse pairs are thin and faint. Inside each bubble is that member's YouTube channel icon, and the colour of the surrounding ring is their own image colour; neither says anything about the relationship (only the centred member's ring is drawn thicker).

Selecting a bubble, or picking a name from the dropdown, switches to a view centred on that member. There, the closer to the centre, the more mentions with that person, and bubble size follows the count with them rather than their overall total. This view exists to show the shape of one member's relationships, which the full map obscures.

The map and the top-30 table hold the same data. If the diagram is hard to read, use the table instead.

The only thing being counted is the number of videos whose title or description names the other member or links their channel. That definition causes specific undercounts and overcounts.

Not counted

• Collaborations that name nobody. A title that says only "collab" without listing participants counts as zero, even when they streamed together.
• Tournaments and events. The title is often just the event name, with teammates left unnamed.
• Streams recorded on one side only. If only one of the two named the other, only that one video enters the total (line weight is the sum of both directions).
• Clips and appearances on other channels. Only each member's own YouTube channel is read.

Counted even though it is not a collaboration

• Mentions in passing. Bringing someone up as a topic, or naming them in an announcement, counts as one.
• Standing link lists in descriptions. Most channels paste a list of agency members into every description, which would make every single video look like a collaboration. To prevent that, the data is split into blocks of 200 videos, and any line appearing in at least half the block is treated as boilerplate and removed. The split matters because these templates change over time: measured across all videos at once, a template used for only part of the period never reaches half, and survives the filter.

On top of that, the number of videos analysed differs per member (26 to 1000). Members with more videos accumulate more mentions in absolute terms, so bubble size cannot be read as a ranking of closeness. Comparisons within a single centred view are less prone to being misread.

Aggregation runs once a day and handles two members per run, with at least six hours before the same member is read again, so a full pass over everyone takes several days. Recent streams therefore do not appear immediately. It is arranged this way to stay inside the free YouTube API quota and the per-request time limit.

Streamer revenue estimator / All tools / Municipalities within a radius