Whoever is responsible for customer service at an online store or a service company rarely chooses channels deliberately. Email was there from the start, chat came with the platform, and the help centre was built once, long ago. The result is often that the same question is answered three different ways depending on where the customer happens to ask it, and that the most expensive channel takes questions the customer would rather have solved alone.
Which questions do customers prefer to solve themselves?
Customers want to solve simple, recurring and time-critical questions on their own, without waiting for anyone. According to Salesforce, 61 percent of customers prefer self-service for simple issues. In Swedish e-commerce these are questions like:
- Where is my parcel, and when will it arrive?
- How do I return, and how long is the return period?
- Can I change the delivery address or cancel?
- Where do I find my invoice or receipt?
- Which size should I choose, and does the accessory fit my model?
What they have in common is that the answer is the same for every customer, or can be looked up in a system without judgement. The problem is not the will but the result. A Gartner survey from 2024 shows that only 14 percent of customer service issues are fully resolved in self-service, and that the figure is only 36 percent even for the issues customers themselves call very simple. Customers start in the help centre, do not find the answer and write an email. Then the company has paid for both channels.
When do customers want a person?
Customers want a person when the issue is complex, emotional, about money or needs an exception, and when self-service has already failed. According to Gartner, 87 percent of customers say it is essential to be able to reach a person when the company uses generative AI in customer service. Salesforce reports at the same time that 81 percent of customer service professionals see phone as a preferred channel for complex issues.
Typical issues where a person is needed:
- Complaints where the customer is unhappy and wants to be heard.
- Issues involving several orders, partial deliveries or incorrect payments.
- Exceptions to policy: a return outside the period, a discount that did not go through.
- Questions that require a decision, for example compensation.
- Anything where the customer has already tried alone and failed.
The last point matters more than it looks. According to a Gartner survey from 2023, 38 percent of Gen Z and millennial customers say they are likely to give up on the issue entirely if it cannot be solved in self-service. They do not email; they buy elsewhere.
What are the strengths and weaknesses of the channels?
Each channel solves one type of issue well and another type badly, which is why no channel can fully replace the others. This is the picture for the three channels, with phone as comparison:
| Channel | Best for | Strength | Weakness |
|---|---|---|---|
| Self-service (help centre, customer portal, chatbot) | Recurring questions with the same answer for everyone, status questions | Immediate answers, around the clock, lowest cost per ticket | Does not solve exceptions; fails silently when the article is missing or out of date |
| Live chat | Questions mid-purchase, short check-ins, handover from a chatbot | The customer stays on the page, one agent can handle several conversations | Needs staffing here and now; poor fit for issues that need documents |
| Email and forms | Issues with attachments, investigation, several steps, sensitive data | The customer does not wait in a queue, the issue is documented, can be assigned | Slow perceived response time, risk of email chains with follow-up questions |
| Phone | Urgent, complex and emotional issues | Fastest at clearing up misunderstandings, feels most personal | Most expensive per ticket, hard to scale, documents nothing by itself |
Two things deserve a comment. Live chat needs someone to be there when the customer is there, and in e-commerce that is often evenings and weekends. And email is not a worse channel than chat; it is a different one: a form that asks the right questions from the start often gives a better first reply than a chat where the agent has to ask their way forward. What customers actually expect in response time per channel is covered in the article on response times in customer service.
What does a ticket cost per channel?
Self-service costs a fraction of a ticket in a staffed channel, but only if self-service actually resolves the issue. The most cited figure comes from a Gartner poll of customer service leaders in 2019: staffed channels such as phone, live chat and email cost on average 8.01 dollars per contact, while self-service on the website and in the app cost around 0.10 dollars.
The figures are US averages from 2019 and say nothing exact about a Swedish company with five people in customer service, but the order of magnitude holds: the difference is manifold. Two caveats:
- The cost is per contact, not per resolved issue. A customer who reads an article, gets no answer and then emails has cost both self-service and email. Low resolution in self-service, as in Gartner's 14 percent, eats the saving.
- The channel's cost depends on what you put in it. A live chat that takes delivery status questions is expensive. The same chat taking questions from customers ready to buy is cheap.
If you want to calculate your own cost per ticket and channel, the method is in the article on customer service metrics.
How do all channels give the same answer?
All channels give the same answer when they fetch it from the same source, a knowledge base with one reviewed article per question and a named owner per article. That is not a technical requirement; it is an organisational one. As long as the help centre is written by marketing, the chat templates by the team lead and the email replies by whoever happens to have the ticket, the answers will differ.
Customers notice. According to Zendesk's CX Trends 2026, 81 percent of consumers want the agent to continue where they left off, and 74 percent are frustrated at having to repeat themselves. Someone who got one answer in the chat and another in the email trusts neither.
This is how it fits together in practice:
- One article, several formats. The returns article is one text. The help centre shows all of it, the chatbot summarises and links, the agent in the inbox inserts the relevant paragraph in the reply. If the return period changes, it changes in one place.
- Channels share history. The customer handed over from chat to email should not have to explain again. The issue, not the channel, is the unit.
- The contact routes are visible. The Swedish Consumer Agency's rules for businesses on customer service say that contact details and how to complain must be clearly stated, and that the company must inform customers if chat conversations are saved. Self-service must never become a way to hide the other channels.
Supportifier is built around that principle: one reviewed knowledge base used in the help centre, forms, chat, inbox and customer portal, so the answer is the same wherever the customer asks. How the parts fit together is described on the platform page. What it takes for a chatbot to stick to that knowledge base and not guess is explained in the article on grounded AI in customer service.
What to do
This can be done in a week without buying anything:
- Pull the last 200 tickets and tag each with channel and type of question. Count how many status and policy questions came by email or chat. That is your self-service potential.
- Search for the ten most common questions in your own help centre, as a customer would. Count how many get a clear answer in the first result.
- Decide per question type which channel should be the first choice and write it down. Delivery status: self-service. Return outside the period: email with a form. Pre-purchase question: chat when staffed, otherwise a form with a same-day reply.
- Compare the answers to the same question in three channels. Take the return period: read the help centre article, the chat template and the three latest email replies. If they differ, you have found your first job.
- Give every common question an owner responsible for the article being correct, and have every channel refer to it.
- Set up the measurement: share of tickets per channel, share that start in self-service and end in a staffed channel, and response time per channel. Follow up in four weeks.
Common questions
Should we drop email if we have live chat?
No. Email and forms are the best channel for issues that need attachments, investigation or several steps, and for customers who cannot sit and wait in a chat. Instead, remove from email the questions the customer could have solved alone, by making the help centre actually answer them. Then the emails become fewer and better, and the response time shorter for the issues that remain.
Is live chat cheaper than phone?
Often, but not always. In 2019 Gartner counted phone, live chat and email in the same cost class, staffed channels, at around 8 dollars per contact on average against 10 cents for self-service. Chat can be cheaper than phone if one agent handles several conversations at once, but it gets expensive if it takes status questions a help centre would have solved. The cost is decided by what you put in the channel, not by the channel itself.
Do we have to offer phone?
The Swedish Consumer Agency's guidance for businesses says contact details and the route to a complaint must be clearly stated, but does not require a specific channel. The Consumer Ombudsman sued a mobile operator in 2025 that only offered a chatbot and hid its phone number; according to the agency's own reporting the court sided with the company. Read the agency's current guidance before you remove a channel, and never make the chatbot the only way in.
Which channel should a chatbot hand over to?
To the channel that is staffed right then and suits the issue. During staffed hours live chat is natural, because the customer is already in the chat. Outside staffed hours a form that collects the order number and the question is better than a chat nobody answers. What matters is that the whole chat history follows, so the customer does not have to start over when a person takes over.
Sources
- Latest Customer Service Statistics To Move Your Business Forward — Salesforce, State of Service, 2025
- Gartner Survey Finds Only 14% of Customer Service Issues Are Fully Resolved in Self-Service — Gartner, 2024
- Gartner Survey Finds 87% of Customers Say Companies Using GenAI for Customer Service Must Provide Access to a Human Agent — Gartner, 2026
- Adapting to the Customer Service Preferences of Gen Z and Millennials — Gartner, 2023
- Gartner Says Only 9% of Customers Report Solving Their Issues Completely via Self-Service — Gartner, 2019
- AI Ushers In Era of Contextual Intelligence, Redefining Customer Experience in 2026 — Zendesk, CX Trends 2026
- Konsumentproblem i kundtjänst: regler för företag — Swedish Consumer Agency, read 2026
- KO stämmer mobiloperatören Hallon — Swedish Consumer Agency, 2025
- Hallon får dölja telefonnummer till kundtjänst — Swedish Consumer Agency, 2026