> Rate the analysishttps://imgur.com/CeLzPgf(For those not about to click a link without some clarification: the above imgur post displays graphs of total watch hours and CCV for Hololive EN, Holostars EN,, Nijisanji EN, VShojo, Phase-Connect EN, and the combined data for Live+StarsEN, going back to September 2022 when VStats first published summary data for NijiHolo.)
Personal takeaways:
Turns out that Ironmouse's subathon really pulled up VShojo's 12-month average for watch hours singlehandedly. Still, the graph from last time suggests that VShojo has still averaged higher CCV in the last six months than the preceding six months that included the subathon, so clearly their new members are getting good retention.
In watch-hour terms VShojo seems to have a real tendency towards monthly variation, one that weirdly tends to inversely correlate with HoloEN for some reason. This isn't reflected on the CCV graph to nearly the same extent, where the trajectory is a lot flatter. Basically, VShojo follows a steadier trend in CCV than HoloEN, but streaming hours seem to fluctuate quite a lot.
This month at least, Phase EN has edged ahead in CCV over both NijiEN and Holostars EN, though it's too early to tell whether this proves to be a permanent inflection point, especially as we don't have data going far back enough to show what the overall trend for Phase is. That said, it does seem likely.
We continue to see a slightly (though not enormously) paradoxical tendency since the start of the year where upward trends in HoloEN's CCV correlate with downward trends in total watch time (itself reflective of lower streaming hours). While my surface-level take is that when streams are less frequent, people are more anxious to watch them, the last time this happened was March when there was 5th Fes hype, so I suspect that the May spike has had similarly one-off rather than structural causes: namely, Advent's Japan arc means less stream time but more hype for said streams. I don't think Gura had a perceptible impact on the May figures, because she streamed twice in April as well. What may well have had an impact, however, was Calli being out for most of the month, as her average monthly CCV tends to be in the 4,500 range (whereas the branch average is usually about 7,500) while she also accounts for about a sixth of the agency's total streaming time in any given month.
I'm sure other, more insightful things are there to be said, but the above are what stood out to me.