Building a Help Centre: structure customers actually use

Key takeaways

• 60% of software users expect a self-service option.
• Always structure categories along the customer journey, not along internal departments.
• One article covers exactly one problem — that prevents endless scrolling and frustration.
• Use the KCS approach: the team maintains the articles while it works on tickets.

Why the structure of the Help Centre is decisive

When users have a question, they look for a quick answer first. If a Help Centre is confusing, they reach straight for the support form or abandon the process. A well-thought-out structure is the key to making self-service actually work.

According to HubSpot, 60% of software users expect a self-service option, and in B2B, 44% start their support request through a self-service channel first¹. Expectations that high call for a logic that explains itself. Anyone who has to search for long feels frustrated.

A well-structured Help Centre filters out recurring standard questions before they land in the inbox as a ticket. That takes lasting load off the support team and frees capacity for complex cases. To get there, making knowledge available has to be planned systematically, so that you can reduce your ticket volume.

• Direct access: answers have to be reachable in two clicks at most.
• Clarity before completeness: plain language beats long documentation.
• Relief for the team: recurring questions are resolved right at the source.

Planning categories and navigation around the user

The most common weak point when building a Help Centre is copying the internal company structure. Categories such as Accounting, Engineering or Marketing mirror the internal organisation, but not the way customers think. Use the customer journey as the foundation instead.

An analysis by HeroThemes shows that 81% of users fall back on knowledge bases and FAQs, and that 91% would use a central knowledge source if it were cut precisely to their concerns². Short procedures and flat hierarchies are decisive here.

Avoid deep nesting. More than two levels of main category and subcategory confuse users. Name categories after concrete actions or problems, for example Getting started, Account & sign-in, or Payment & invoices.

Make sure the navigation works on every device. Many searches happen on mobile. An overloaded sidebar makes content harder to find on small screens.

Article scope: one problem per article

A common mistake when writing support content is the endless wall of text. If a single article covers setup, billing and troubleshooting all at once, readers scan it in vain for their specific question. The principle of modularity is the remedy.

Structure content by this rule: an article covers exactly one problem. If a topic needs several steps, create separate pages and link them to each other. That makes the whole thing easier to survey and easier to navigate.

• A concrete title: name the exact problem in the main title.
• Short paragraphs: keep paragraphs between 80 and 150 words for readability.
• A focused solution: drop the introductory filler and get to the point.

Modular content is also easier to maintain later. If a feature changes, you only adjust one short article instead of reworking a long manual. Building a knowledge base like that requires clear rules for the whole team.

Optimising searchability and metadata

Most customers use the Help Centre's internal search instead of clicking through menus. If the search returns no matching hits, users abandon the process. Searchability stands or falls with the language of the articles.

Avoid internal jargon and abbreviations only your team knows. In the KCS approach, the support team captures the solution inside the step of handling the request and records the context of the person asking, that is, their words and phrasings; where no matching article exists yet, exactly those search terms become the starting point of a new one³. That is precisely why the customers' own words end up in the knowledge base automatically.

• Add synonyms: record alternative terms such as password, passcode and login as tags.
• Choose precise titles: use solution-oriented titles such as Reset your password instead of Account security.
• Maintain metadata: put the relevant keywords into the article tags, for the search engine.

Check regularly which search terms people enter. If customers search for exchange but your article only talks about returns, the search leads nowhere. Adjust the metadata accordingly.

Maintaining knowledge day to day (KCS approach)

A Help Centre is not a one-off project. Products change, new features arrive and existing processes shift. If the Help Centre is not updated continuously, the knowledge in it ages fast.

The Knowledge-Centered Service (KCS) approach solves this problem by letting knowledge emerge as a direct by-product of daily ticket work. Atlassian describes KCS like this: support teams do not only provide support in real time, they create and maintain the documentation in the same work process. Whoever handles a request checks the knowledge base first, updates an existing article where steps are outdated or unclear, and documents the solution anew when no article exists yet⁴. So instead of running costly annual audits, the support team adjusts articles right while answering requests.

• Step 1: capture. Record new questions from the ticket straight away as a draft article.
• Step 2: structure. Use article templates for consistent formats.
• Step 3: reuse. Link existing articles in ticket replies.
• Step 4: improve. Correct wrong or outdated paragraphs immediately.

With this method the documentation stays current without extra work. The support team takes joint responsibility for the knowledge base instead of delegating upkeep to individuals.

Measurement and continuous optimisation

Without data it stays unclear whether your Help Centre brings the relief you wanted. Analysing user behaviour shows you exactly where articles are missing or badly worded.

Track the three figures Ferndesk names as a scorecard for good self-service: searches without a hit as a sign of content gaps, articles with many views and negative feedback as a sign of quality problems, and the path from self-service into a ticket⁵. Watching these metrics protects against a loss of quality.

Evaluate the results of the zero-result searches weekly. That way you spot new problem areas or changed terms after a product update early.

Technology and AI: using knowledge twice

A modern Help Centre is no longer just a place for human readers to turn to. It is at the same time the structured basis for automated support systems. If your knowledge is cleanly prepared, you can use it for intelligent assistants without extra work.

Prevent redundant data storage. The content of your Help Centre should serve as the central knowledge base (single source of truth) that widgets and chatbots draw on as well. That way you only have to enter a change in one place.

Platforms such as ComLayer use the articles in your Help Centre to answer routine questions in the chat widget automatically and with an exact source citation. Cleanly structured self-service and modern AI support thus combine into an efficient support architecture with which you can automate your customer service.

Frequently asked questions