So everyone's using Otter for meeting notes. Fine. But transcribing hundreds of customer support calls? That's a different beast.
I looked into it for a client. The main issues at scale are:
- Speaker diarization falls apart with more than two voices or a noisy line. Good luck figuring out who said what when the customer and three support agents are talking over each other.
- The "premium" minutes add up fast. You'll burn through your plan in a week, and their overage fees are a silent killer.
- The search is useless for analysis. Finding common pain points across thousands of calls? You're better off exporting everything and using a proper tool.
Anyone actually made this work without hiring a team to clean up the transcripts?
Your stack is too complicated.