@azukichannel3 Channel Audit: 35.7K Subs in Japan's AI Vtuber Niche

Quick honesty note before anything else: the live scrape pulled 30 long-form uploads from the last batch with blank titles and 0 views attached to each row. That's almost certainly a scraping artifact (private playlist, age-gating on the feed, or rate-limit timing) rather than 30 actual 0-view uploads — a channel sitting at 35.7K subs with 204 lifetime videos doesn't realistically post 30 ghost videos. So this audit leans on the structural data that did come through cleanly: subs, total views, video count, niche positioning, country, and the upload-cadence claims in the channel description.

Lifetime math first. 2,554,981 views over 204 videos is ~12,524 views per video averaged across the entire catalog. That's a respectable number for a Japanese AI/creative Vtuber — the niche is real but narrow, and 12K average means at least a decent chunk of the back catalog is doing work, not just one breakout viral hit carrying the rest. 35,700 subs against 204 uploads gives a subs-per-video ratio of ~175, which suggests the channel converts viewers into subscribers at a healthier rate than typical AI tutorial channels (which tend to skew low-sub because viewers want the answer and leave). The Vtuber wrapper is probably doing that work — character-led channels build parasocial subscriber loyalty in a way faceless tutorial channels don't.

The positioning is the most interesting thing here. "AI×クリエイティブを発信するVtuberあずき" — AI creative tools, taught by a Vtuber character, with the creator also working as an illustrator. That's three layers stacked: AI tooling (music gen, video gen, image gen per the description), Vtuber persona, and illustrator credibility. Each layer narrows the audience but deepens trust within it. The Japanese AI-Vtuber crossover is small enough that 35.7K is meaningful share-of-voice, not a rounding error.

Now the gap. Content mix on the last 30 uploads is 0 Shorts, 30 long-form. Zero. In 2026, a channel with weekly long-form and zero Shorts in the recent 30 is leaving the largest discovery surface on YouTube untouched. The Sunday 18:00 weekly AI news slot is the obvious raw material — every news item is a 30-60 second Short waiting to happen, and AI-news Shorts in Japanese have weak competition compared to the English-language saturation. Even a 1-Short-per-news-item habit would test whether the algorithm pushes the character into a wider feed. Worth saying out loud: I can't see whether Shorts were tried earlier and pulled because they underperformed, so this is a hypothesis, not a verdict.

The weekly news cadence (Sunday 18:00) plus irregular Friday deep-dives is a smart format split on paper — news for retention/repeat-viewer habit, deep-dives for SEO/long-tail discovery. The risk is the deep-dives being "irregular." From outside the channel, irregular publishing on the discovery-focused format is usually what caps a channel's growth ceiling, because the algorithm rewards predictability on the format that pulls new viewers in. If the Friday deep-dives moved to a fixed cadence (even bi-weekly), the search-driven tail on those would likely compound faster than the news slot.

One thing I genuinely don't know from outside: retention curves, CTR, or which specific tools (Suno, Midjourney, Runway, KlingAI, etc.) are driving the strongest views. The illustrator background plus AI generation focus suggests the channel could own a defensible niche around "AI tools that respect the illustrator workflow" — that's a sharper hook than generic "AI for creators," and it's a positioning the data hints at but doesn't confirm.

If I were the channel owner, the move I'd test in the next 90 days is small: pick the single highest-performing video theme from the last 6 months and make three deep-dive followups on adjacent tools. The 12.5K lifetime average suggests the catalog already has signal — finding which video over-indexes and doubling down on its angle usually does more than chasing a new format.