Customer service
What is customer service and where does it end?
Customer service is answering questions your own setup caused. That sounds harsh and it's the most usable definition there is.
In short
Customer service is the whole of the work by which a company answers and handles questions, complaints and requests from customers, across every channel where customers make contact. It concerns contact the customer initiates, which distinguishes it from sales and marketing, where the company starts.
Where do your questions come from?
Every customer question has an origin outside customer service. A question about delivery time comes from a shipping page that doesn't say. A question about a return comes from a policy stated differently in two places. A question about a product comes from a product page that omits sizing.
That makes the department the place where the rest of the company's mistakes become visible. Grouping tickets by topic therefore gives you the sharpest list of improvements for the website, the logistics process and the product information that exists anywhere in the company.
It's also why a service team judged only on response time never gets less work. Answering the same question faster changes nothing about how often it's asked.
The three kinds of work that get mixed up
What a service team does in a day falls into three kinds that need different things. Throwing them into one list is the most common cause of busyness nobody can explain.
- Information questions
- Where's my parcel, what's the delivery time, does this product fit that one. The answer already exists somewhere. This is the part AI and a good website can genuinely remove.
- Actions
- Change an address, register a return, cancel an order. The answer isn't the point; something has to happen in a system. Automating here means letting the customer do it themselves, or putting the action behind one click.
- Judgements
- A complaint, an exception, a goodwill request. Here someone has to decide, and that's the work you keep people for. Everything you remove from the first two kinds frees up here.
How many people do you need?
The calculation isn't hard, but it's often done on the wrong denominator. Productive time isn't eight hours. Meetings, breaks, training and admin take a substantial share out in practice, and peak hours don't spread evenly across the day.
More important than the number is doing it per channel. Chat needs simultaneous staffing at fixed hours, email can be spread across the day, phone needs someone free at that moment. An average across all channels hides exactly the hour where it breaks.
What AI changes about the work
The expectation is usually that AI lowers contact volume. What happens first in practice is different: the composition shifts. Information questions get handled and what's left is heavier, because those are the judgements and exceptions.
That has two consequences that surprise teams. Average handling time rises, because the easy conversations are gone. And the skills you need change, because more of the remaining work requires someone to make a decision and explain it.
Where it goes wrong
Steering on speed without looking at repeats
A team answering the same question faster is working harder on the same problem. Alongside response time, measure how often the same customer comes back about the same thing, because that number points at the cause.
Giving the department no mandate to change anything
Customer service sees first which page is unclear, and usually can't do anything about it. Without a fixed route to the people who own the website or the process, that knowledge stays inside the tickets.
Opening every channel without being able to staff them
A WhatsApp number that replies after two days is worse than no WhatsApp number. Pick the channels you can honour on your busiest day, not on a quiet Tuesday.
Treating customer service as a cost centre
The department that hears first that a product structurally underperforms isn't a cost centre but your earliest signal. If you steer only on cost per contact, that signal disappears into handling time.
Reducing your questions in four steps
This order works better than automating straight away, because otherwise you answer questions faster that never needed asking.
Group a month of conversations by topic
Not by channel and not by agent. Topic is the only grouping that points at a cause.
For the biggest category, find where the customer looked
There's usually a page that should have given the answer. Changing that page removes questions instead of answering them faster.
Turn the second category into a self-service action
Change an address, register a return, request an invoice. These are actions, not questions, and a customer would rather do them than wait for you.
Point AI at the information questions that remain
Only now. What you removed before doesn't need answering by AI, and the answers it gives come from a source that's now correct.
What to do this week
Take a hundred conversations from last month and list them by topic. You don't need more for a distribution that holds. Then check for the largest category whether the answer is on your site, and if so, where the customer searched before contacting you. That's the cheapest improvement customer contact has to offer.
What this looks like in a system
In Cuego email, chat, phone and WhatsApp arrive in the same list, with the customer and their orders alongside. Topic and urgency are determined on arrival, so the breakdown by topic exists immediately instead of being counted afterwards. The AI agent handles information questions within the boundaries you set, and what remains reaches an agent with the history attached. From a conversation you start a task, a return or an appointment.
Automation & workflowsRelated terms
Contact ratio
Contact ratio is the number of customer questions per order. It's the only number you can't improve by working faster.
Read onWISMO
WISMO is the question of where the order is. It's rarely a location question and nearly always a certainty question, and that difference determines what you can do about it.
Read onHelpdesk
Helpdesk is the collective term for the desk where questions and reports arrive. The term covers both a team and the software, and that confuses matters when choosing.
Read on