The same twenty questions, answered from scratch
Support arrives in a shared inbox with no owner, no priority and no service level. The same questions get researched and answered again every week, and retention risk only becomes visible after the customer has decided.
- Tickets per month
- About 700
- Routine questions
- Around 60 percent
- Hours released per month
- About 60
- Escalations
- With full history attached
The situation
How to read this. The figures below are worked estimates, not a client result. They are modelled on the volumes a typical operational SME runs at, and the arithmetic is shown so you can put your own numbers in. We size the real case against your actual volumes before quoting.
Around 700 support requests a month land in a shared inbox across email and WhatsApp. A large share are the same small set of questions: order status, document requests, opening hours, how to book, what is covered. Each is answered from scratch by whoever picks it up.
Because nothing is structured, nobody can say how long resolution takes, which customers raise the most issues, or which problems keep recurring.
What we build
Requests are classified, prioritised and assigned an owner automatically, with the customer record attached. Routine questions are answered from an approved knowledge base on controlled channels, transparently, so nobody is pretending to be a person. Anything outside that scope routes to a human with the full history so the customer does not repeat themselves.
The numbers that move
- Hours saved. If 60 percent of 700 monthly tickets are routine and each takes around nine minutes end to end, that is roughly 60 hours a month returned to the team.
- Response time. Routine answers become immediate, at any hour, which is where most of the perceived service improvement comes from.
- Retention signal. Issue patterns per customer become visible while the relationship is still fixable.
- Measurable service. Response and resolution times exist as numbers for the first time.
Common questions
Yes, and they should. We do not build systems that pretend to be a person. The agent handles approved questions transparently and hands over the moment a case needs judgement.
Where to go from here
Tell us what you'd like to do with AI.
Plain English is fine. We'll ask the technical questions on the call, and if we're not the right fit we'll say so on the first one.