Sierra uses support volume as moat
Sierra
Sierra is using support volume as a training advantage, not just a revenue stream. Traditional consumer brands generate huge numbers of repetitive questions about returns, billing, subscriptions, shipping, device setup, and account changes, which gives Sierra more real conversations to tune workflows, escalation rules, and action taking into back end systems. That matters because in AI support, the hard part is not answering one question, it is resolving millions of similar ones reliably in a brand safe way.
-
Sierra’s customers skew toward high volume consumer support environments like WeightWatchers, SiriusXM, Sonos, and ADT, while Decagon’s base has tilted more tech centric with customers like Notion, Rippling, and Duolingo. That gives Sierra denser exposure to the repetitive, operationally messy cases where resolution quality improves fastest with scale.
-
The economic loop is unusually strong. AI support agents are priced around $0.99 to $1.50 per resolution versus roughly $10 to $15 for human handled tickets, and third generation agents can resolve 60% to 80% of conversations without a person. More resolved volume means more savings for customers and more production data for Sierra.
-
Sierra’s white glove deployment model helps it capture the highest value data, because engineers wire the agent into CRM, billing, ERP, and help desk systems so the agent can actually process refunds, update subscriptions, and complete tasks. That creates learning around full workflows, not just chat transcripts, which is harder for lighter weight chatbot vendors to match.
If this continues, the winners in AI customer service will look less like software widgets and more like scaled operators with proprietary interaction data across chat and voice. Sierra’s move into phone based support should accelerate that advantage, because voice still accounts for most customer service volume and expands the dataset from typed questions to full end to end service conversations.