I've been a longtime Rev user for client interview transcriptions, but my team was getting tired of the 24-hour turnaround. The promise of Otter.ai's real-time transcription and its slick integrations felt like a game-changer for our remote agency. So, I made the switch for a few key projects last month.
Honestly, I regret it for anything that goes to a client. The accuracy just isn't there for the nuanced, jargon-heavy conversations we have. With Rev, I got a human-polished transcript where industry-specific terms and client product names were correctly captured. Otter.ai turned a key technical discussion about "API webhooks" into "AP eye web hawks," which is just embarrassing to have to clean up before sending. The time I saved on the initial turnaround was completely eaten up by my own extensive proofreading and correction.
It's fantastic for internal brainstorming sessions or casual team meetings where you just need the gist. The live notes feature is genuinely cool! But for a deliverable? The confidence isn't there. I found myself second-guessing every proper noun, which defeated the purpose.
I'm switching back to Rev for all client-facing work. The peace of mind is worth the wait and the higher per-minute cost. Maybe I'll keep Otter for internal syncs, but it's a lesson learned: "good enough" AI isn't always good enough when your reputation is on the line. Has anyone else hit this wall with Otter for professional, client-ready output? Curious if I'm just using it wrong.
Hey there, totally feel your pain. I run the support ops for a 12-person SaaS dev shop, and transcripts for client calls and product feedback sessions are mission-critical for us. We've lived with Otter (Business) for over a year, tried Rev once for a flagship client project, and we also use Descript for some post-processing. Here's my breakdown.
1. **Cost and Turnaround Structure**: Rev is a per-minute service, typically around $1.50-$1.80 per minute for their human transcription with 24-hour service. Otter is a subscription - we pay $30/user/month billed annually. Otter's cost is effectively infinite minutes, but you're paying with your time. The hidden cost of Otter is the 10-20 minutes of intense proofreading per hour of nuanced technical talk.
2. **Accuracy & Context Window**: Otter's engine is general-purpose and struggles, as you noted, with jargon and proper nouns. In my environment, for a call about "Kubernetes cluster autoscaling," I've seen "cooperate this cluster auto singing." Rev's human transcribers, when given a glossary (which you can upload), nail that stuff. A concrete detail: Otter's accuracy on our technical calls feels about 85-90% for the gist, but Rev's is genuinely 99%+ for a polished deliverable.
3. **Integration & Workflow Ops**: Otter wins on live integration. The Otter Bot joining Zoom calls and providing live, searchable notes is a genuine productivity boost for internal standups. Rev is a manual upload-and-wait process. For post-call workflows, Otter's direct Google Docs export is nice, but if you have to re-listen to the whole thing to fix errors, that advantage disappears.
4. **Vendor Lock-in and Data Handling**: This is a quiet one. With Otter, all your conversation history is in their cloud. With Rev, you get the final transcript file and that's it. If you ever cancel Otter, accessing your old transcript search is a pain. For client work involving sensitive IP, I prefer Rev's model - it's a transaction, not a repository.
My pick? I recommend Otter for **internal, non-critical meetings** where speed and search are key. For **any client-facing deliverable or a recorded conversation that becomes a source of truth**, I'd go with Rev every single time. The peace of mind is worth the per-minute cost. To make a clean call, tell us if your client calls are under NDA and what your average monthly transcription minutes actually are.
That's a really specific example, and it clicks for me. I've been using Otter for some internal notes, and I see the same thing with technical terms. It struggles hard with acronyms.
Do you think Rev's human editors get better because they can look up a client's brand guide or something? Or is it just the human ear being smarter about context?
rookie
Yeah, the "API webhooks" to "AP eye web hawks" is painfully relatable. It's that last-mile accuracy for proper nouns and jargon where the automated tools still stumble.
For anything internal, the speed is awesome, but I've found the same thing - once you're spending more time proofreading than you'd spend just waiting for a human service, the math falls apart. The real-time aspect feels like a luxury until it becomes a liability because you're now the editor.
Have you looked at any of the middle-ground options? Like using Otter for the live draft and a lightweight human pass from a service like Scribie just for the final polish? Might keep some speed while recovering that client-ready accuracy.
System calls per second matter.