What 784 Support Emails Tell You About Ecommerce Support
We analyzed 30 days of customer support emails from one ecommerce store. 784 replies in total.
This is how the cases broke down:
- Shipping issues: 29%
- Complaints: 18%
- Order changes: 17%
- Refund requests: 5%
Two things stand out. Shipping and logistics dominate the queue — nearly a third of all tickets. Refunds, on the other hand, are rare but slow: agents need to check order status, verify returns, and match the store’s refund policy before sending anything. One wrong reply costs more than ten routine ones.
The more actionable split was this:
- 64% of emails were straightforward enough for AI to draft a full reply
- 35% needed an AI draft with human review before sending
- About 1% required fully manual handling
That 35% is where most tools get the framing wrong. The goal is not to automate those emails — it is to give the agent a starting point that already knows the order details, the customer history, and how the store has handled similar cases before. Review takes seconds instead of minutes.
This is why we think the future of AI in customer support is decision support, not full automation. The 1% that needs full human attention will always exist. What changes is how long the other 99% takes.
Sensbot builds this context from a store’s past support emails. Before generating a reply, it maps how the store handles refunds, complaints, and order changes in practice, not just what the FAQ says. That context travels with every draft.
The result is fewer replies that need to be rewritten, and a shorter queue that needs human time.
See it in action