GLOSSARY

Customer self-service and how to make it work

Customer self-service is every way customers find an answer or fix a problem without contacting you, from a help centre to a billing portal or an AI chat.

It comes in two kinds. Customers look something up, in your docs, an FAQ, a community forum or a chat that answers from your docs. Or they do something themselves, such as changing plan, updating a card or resetting a password, in their account settings or a billing portal. Either way, the question never reaches your inbox.

Most customers try it before they write to you, and most don’t get all the way. In Gartner’s survey of 5,728 customers in December 2023, 73% used self-service at some point while getting help, but only 14% of issues were fully resolved there.

For a founder answering support alone, self-service is the difference between answering a question once and answering it every week. Deacon is an AI customer support agent that answers your customers on your site in seconds, from your own docs, and it has a free plan.

The kinds of customer self-service

Each kind suits different questions. For a small SaaS, docs and a billing portal come first.

Docs or help centre

What customers do there
Read articles, often found through search
Best for
Setup steps, how-tos, anything that needs a screenshot

FAQ page

What customers do there
Scan short answers on one page
Best for
The ten questions every new customer asks

Community forum

What customers do there
Ask other customers
Best for
Products with keen power users, once there are enough of them

AI chat

What customers do there
Ask in their own words and get an answer from your docs
Best for
Customers who won’t search, and questions at 3am

Account settings and billing portal

What customers do there
Change things themselves
Best for
Plans, cards, invoices, cancelling and passwords

Status page

What customers do there
Check whether something’s down
Best for
Outages, so a hundred people don’t email about the same one

The doing kind is often the quickest to set up. If you bill with Stripe, its customer portal lets customers update their card and billing details, change or cancel a subscription, and pay, view and download invoices. You switch it on in the Stripe Dashboard without writing code, and customers can fetch last month’s invoice without emailing you for it.

How to measure customer self-service

Two rates answer two different questions. The first says how well self-service works for the customers who try it. The second says how much of your support it carries.

Self-service resolution rate = customers who solved their problem in self-service ÷ customers who tried self-service × 100

Self-service rate = questions answered in self-service ÷ (questions answered in self-service + questions that reached your team) × 100

Say 500 customers used your docs or chat in June. 350 found their answer and never wrote in, and 150 emailed you anyway. Another 100 emailed without trying self-service at all. Your self-service resolution rate was 350 ÷ 500, or 70%. Your self-service rate was 350 ÷ (350 + 250), or about 58%.

The second is what most support tools call deflection rate, and that entry covers how to count it honestly. The hard part of both is knowing who solved their problem. A customer who gives up and closes the tab looks the same as one who found the answer, unless you ask. A one-line “Did this answer your question?” under each article gives you a count, and matching a week’s emails to that week’s chats catches the people who wrote in anyway.

What the research says about self-service

Customers want it, and most of it doesn’t work for them yet. These are the published figures we found, read on 23 September 2026.

Customers who used self-service at some point while getting help

Sample
5,728 customers surveyed by Gartner, December 2023
Figure
73%

Issues fully resolved in self-service

Sample
The same
Figure
14%

Customers who started in self-service and said the company didn’t understand what they were trying to do

Sample
The same
Figure
45%

Customers who’d rather use self-service for simple issues

Sample
Salesforce’s State of the Connected Customer, August 2023, as quoted in its State of Service report
Figure
61%

Customer issues self-service solves, as estimated by service professionals at organisations that offer it

Sample
More than 5,500 service professionals in 30 countries, December 2023 to January 2024
Figure
54%

Consumers highly satisfied with the self-service tools they’re given

Sample
1,320 respondents in the US and the UK, for NICE, published May 2022
Figure
15%

The share of consumers businesses believed were very satisfied

Sample
The same
Figure
53%

The gap between the last two rows is the one to worry about, because businesses think their self-service works far better than customers do. Salesforce asked agents at organisations with self-service why it falls short. 65% said customers’ issues are often too unique for it, 62% that materials aren’t always up to date, and 59% that instructions are often too complicated to follow. In Gartner’s survey, the most common reason self-service failed, in 43% of cases, was that customers couldn’t find content relevant to their issue.

Making self-service work for a small team

Gartner’s advice to large support teams, in the same release, mostly scales down. Its first fix is to have the people who answer questions write the missing content as part of answering them, rather than as a separate job. It also recommends one clear place to start on your website, such as an AI chatbot, and a way for customers and your team to flag content that didn’t help. For a founder, that comes down to four habits.

  • Answer each repeated question in your docs once, the day it comes up a second time.
  • Put a billing portal behind every “change my plan” and “send me my invoice” email.
  • Put answers where customers ask, not only where they might search.
  • Keep a way to reach you in plain sight. Hiding your email raises your self-service rate and loses you customers, and some messages should always reach a person.

Self-service vs knowledge base, customer portal and deflection

These get used as if they were one thing. Each is a different part of it.

Customer self-service

What it is
Every way customers help themselves, the whole idea

Knowledge base

What it is
The articles self-service answers from, whether customers read them or a chatbot does

Customer portal

What it is
The signed-in part of your site where customers see their account, invoices or past requests

Deflection rate

What it is
The share of questions self-service answered before they reached you

Containment rate

What it is
The share of conversations a bot finished without handing them to a person

What self-service can’t answer lands in your inbox, or in a ticketing system once a team shares the work. Every question it answers is one fewer in either.

Self-service where your customers ask, with Deacon

Deacon takes the searching out of self-service. Your customers ask in their own words, in a chat on your site, and Deacon answers in seconds from your docs, at any hour and in the language they wrote in. That’s aimed at the most common failure in Gartner’s survey, customers who couldn’t find the right content, because nobody has to find it.

It gives them a way through to you. When your docs don’t cover a question, Deacon says it doesn’t know instead of guessing and asks for the visitor’s email, so you can reply yourself. The owners on your team get an email when a visitor leaves an address or asks for a person.

It works the way Gartner’s first fix describes, with the missing content written as you answer. Every question your docs didn’t cover waits in Conversations with an Answer this button. What you write becomes part of what Deacon knows, so it can answer the next person who asks, and Save & check shows the reply Deacon now gives and how many other open questions your answer covers. The more you answer, the less it needs you. FAQ chatbot walks through the flow, and knowledge base chatbot covers what Deacon reads.

It shows you where your self-service falls short. Topics group similar questions, count the people who asked and label most groups as a how-to, a bug report, a feature ask and so on, so you see what customers are stuck on and what they want next. Analytics › Widget names the page where people get stuck, once one page’s questions go unanswered more often than anywhere else.

Two limits are worth knowing. Deacon always declines questions about a visitor’s own account, order or data, and it takes no actions, so keep a billing portal for the doing kind of self-service. And the widget has no rating on an answer, so Deacon can’t tell you a self-service resolution rate. The Answered figure on Analytics › Widget is the share of questions Deacon answered out of those it tried, and it counts greetings and requests for a person as answered. You can try it all on the free plan, which needs no card.

Try it on your own documentation

Add your help pages, ask it the question you know your docs cannot answer, and watch it say so.

Start free

Free plan, no card.