Role
Customer service
Which ticket comes back most often? If the team can name five from memory, each of them is being answered for the thousandth time - from scratch every time and slightly differently every time.
Intro call
Four questions we would start with
Which kind of ticket repeats most often?
Usually a handful of topics account for most of the traffic. Each has an answer that does not change between clients - and yet it is written afresh with every ticket.
How quickly does a client get a first reply?
For a ticket raised on a Friday afternoon the reply time is sometimes longer than the matter itself. The client judges the service by that time before they read a word of it.
Where does the knowledge from a closed case go?
If it goes into the ticket history and nowhere else, the next agent solves the same problem from zero - and the client gets a different answer than last time.
How many times does a client have to repeat the same thing before a case is resolved?
Every handover between people is a repetition of the story. That is the part of service clients rate worst, whatever the final outcome.
We take over the searching, you keep the resolving
Recognising a ticket's topic, collecting the client's context and preparing answers to typical matters can be taken off the desk. The conversation with the client and the decision on a resolution stay with a person.
- A prepared answer is not a sent answer. - The agent edits it and the agent sends it. Automatic sending to a client is a line we do not cross in this role without an explicit decision.
- The client's context before the answer. - What they bought, what they raised before, whether they are waiting for something else. An answer given without that is often correct and useless.
- A closed case feeds the knowledge base. - The resolution goes where the next person will find it - and that is why the client gets the same answer whoever picks the case up.
Scope
What can be taken off customer service
Nineteen example tasks. Each one has a human approval point. Below is a selection, and the scope we fit to your process.
Taking in a ticket
- Collecting tickets from every channel into one place
- Recognising the topic and assigning it to the right queue (verified by: the agent)
- Marking urgent tickets and key clients' matters
- Detecting a ticket about a case that is already open
- Acknowledgement with a realistic reply time
The client's context
- The history of purchases, tickets and contacts before the answer
- Pointing out this client's open matters in other departments
- The status of an order or a service
- What was agreed with the client before
Resolving the case
- A proposed answer for typical topics (edited and sent by: the agent)
- Pointing out the procedure that applies
- Preparing the documents needed for a resolution
- The list of steps for multi-stage matters
- A reminder about cases waiting for the client to act
Escalation and knowledge
- Marking cases that need passing to a specialist (decision: the agent)
- The full picture passed on together with the case
- Recording the resolution in the knowledge base (approved by: the agent)
- Pointing out topics with no ready answer in the base
- A summary of the causes of tickets - material for other departments
A day's work
Today
- You collect tickets from several inboxes and channels
- Before answering you check the client's history in three systems
- You write the same answer for the twentieth time this month
- When a case is handed over the client repeats everything again
- The resolution of an unusual case stays in your head
- At the end of the day you do not know which cases are waiting on the client
With AI assistance
- The tickets wait in one place, with urgency marked
- You have the client's context before you open the reply
- For a typical topic you get ready material to edit
- A case handed on goes with the full picture
- The resolution goes into the base and serves the next person
- Cases waiting on the client have their own list with dates
- Time to first reply
the client does not wait until Monday
tickets are recognised and queued straight away
- Resolved first time
fewer handovers between people
the agent has the context and the procedure at hand
- One version of the answer
the client gets the same answer whoever handles the case
the resolutions come from the base, not from memory
- Turnover in the team
the work stops being retyping the same thing
the repetition falls and what remains is resolving
We start with one channel and the few most frequent topics.
- Step 1
We count the tickets
Split by topic and channel. Usually a handful of topics account for most of the traffic - and that is where we start.
- Step 2
The client's context in one place
Purchase and ticket history before the answer. A simple change, felt straight away.
- Step 3
Material for answers on typical topics
With full editing on the agent's side. We measure how much has to be corrected.
- Step 4
The knowledge base and the lessons
Recording resolutions and summarising the causes of tickets for other departments. Last, because it needs history.
Common questions
Questions that come up at the first meeting
Is this not plainly a question about cuts in service?
It is, and we answer it honestly: we do not promise cuts and we do not build the argument on them. In customer service the volume grows with the company and the team usually does not - and that is the situation this is most often discussed in. A team that is not hunting for context before every answer handles more cases and handles them better. What follows from that for headcount is your decision and is made independently of us.
Does this fall under the AI Act?
Handling tickets is not a high-risk use. The transparency obligation concerns a situation where the client is talking to an automation - in our approach they are not, because a person sends the answer.
Still have a question that is not here? Book a call
Next step
How many times this week has your team written the same answer?
We will show you what service looks like when the client's context is waiting ready and the resolution of a closed case serves the next one.