Give Okou a video link. It reads the public comment thread and returns what the audience is doing there — the themes, the requests hidden inside praise, and the quotes worth pulling — with the sampling limits stated.
Comment analysis is not sentiment scoring. On an appreciative thread every score is positive and the number tells the channel nothing it can act on.
What a thread actually contains is behaviour: people summarising the video because it has no chapters, quoting one sentence because it is the clip, or writing in another language because there are no subtitles. Reading for that is what turns a comment section into a backlog.
Four things a view count cannot.
What the audience is doing in the thread, grouped by behaviour rather than by positive and negative.
The workaround a viewer built because something was missing, which is a stronger signal than a request.
The lines the audience already repeats back, which is clip selection done for you.
Traces of a traffic source or a change in who is arriving, visible in the thread before it is visible anywhere else.
A thread full of hand-written timestamps is a production brief. The workflow below takes the same source material through to a finished asset.
A real public comment thread read three ways, with the sampling bias stated and commenter handles left out.
Six behaviour-based clusters across 40 real comments, with the finding that five viewers hand-built the video's missing timestamps.
Five requests ranked by evidence, including the ones nobody phrased as a request, plus the chapter list taken straight from the viewers.
The comment arrival curve across six years, what an evergreen shape changes about maintenance, and why the default sort understates recent volume.
Connect the channel and let Okou read the comments on each new upload, cluster them the same way every time, and send you only what changed.
YouTube
Notion
Slack
Google SheetsConecte os aplicativos que contêm o material de origem mais recente. Reutilize as etapas como um workflow. Adicione uma automação apenas quando o workflow precisar começar por um gatilho ou agendamento.
As many as the thread exposes, in the order the platform sorts them. The output states how many were read and how the sort biases the sample, rather than implying a census.
No. Comment text is quoted because it is public; handles are left out because they add nothing to the analysis.
Yes. Give it both links and say what you are comparing — themes, requests, or how the audience changed between them.
O Okou também tem: youTube channel audit e video to blog post.
Connect the channel and let Okou read the comments on each new upload, cluster them the same way every time, and send you only what changed.