TL;DR
- The single most common mistake: organizing by product feature instead of by user task. Customers do not know your feature names; they know what they want to do.
- Three methods work: organize by user task, by customer journey stage, or by product/service. Pick one method, do not mix them.
- Restructuring without losing SEO is doable: use 301 redirects, keep article URLs stable, only the category structure changes.
The 3 most common knowledge base structural mistakes, in order of how often they tank usability:
- Categories named after product features ("Webhooks", "API v2", "OAuth Configuration") instead of user tasks ("Connect external tools", "Authenticate API requests").
- Too many top-level categories. Seven or more categories means customers cannot scan the list and have to search instead, which means they leave.
- Inconsistent depth. Some categories have 2 articles, others have 30. The 2-article categories should be merged or deleted; the 30-article categories should be split.
This guide covers three organization methods that actually work, with concrete examples for SaaS products, plus the migration path if your existing structure has these problems.
Method 1: Organize by user task
The default method for most SaaS help centers. Categories map to what the customer is trying to do, not to your product anatomy.
Sample structure for a SaaS analytics tool:
- Getting started, first-week setup tasks
- Connect your data, integrations, imports, syncing
- Build dashboards, creation and customization tasks
- Share and collaborate, team, permissions, exports
- Analyze and report, querying, alerts, scheduled reports
- Manage your account, billing, plan changes, security
Why it works: customers think in tasks ("I want to share a dashboard"), not features ("I need to use the share API endpoint"). Task-based categories match the search query language customers actually use, which means better SEO too.
Trade-off: when one task touches multiple product areas, you have to choose where the article lives. Pick the primary task, then link from related tasks. Example: a webhook setup article lives under "Connect your data" but is also linked from "Build dashboards" because dashboard alerts use webhooks.
Method 2: Organize by customer journey stage
Used by tools where the customer experience is sequential and onboarding-heavy.
Sample structure for an HR onboarding SaaS:
- Before you sign up, eval and trial questions
- First week setup, initial configuration
- Onboarding your team, getting users in and trained
- Daily operations, recurring workflows
- Scale and optimize, power features for growing teams
- Account and billing, recurring admin
Why it works: customers know roughly where they are in their journey, so categories match their mental state. Particularly strong for products with a clear onboarding arc.
Trade-off: later-stage articles get less traffic but more impact (advanced users matter). Avoid the temptation to write less for them.
Method 3: Organize by product or service
Best for multi-product companies or platforms where each product has a distinct user base.
Sample structure for a marketing platform:
- Email campaigns, everything email
- Landing pages, page builder
- Forms and surveys, data collection
- Automation workflows, triggers and sequences
- Analytics and reporting, measurement
- Account and billing, admin
Why it works: customers know which product they bought. They do not need to learn your taxonomy.
Trade-off: when multiple products share concepts (segments, contacts, branding), the same article ends up duplicated under each product or referenced indirectly, which wastes content effort.
Naming conventions for categories
Three rules that hold across all three methods:
Rule 1: Use 2 to 3 word category names. "Account & Billing" is good. "Account, Billing & Subscription Management Settings" is bad. Long names crowd the navigation and read like internal taxonomy meetings.
Rule 2: Lead with the verb where possible. "Connect your data" is stronger than "Data connections". Verbs match customer intent.
Rule 3: Avoid your internal product names in categories. "Workspace settings" beats "Studio settings" if "Studio" is your internal name for the workspace concept. Customers do not know your internal vocabulary.
Use analytics to spot structural problems
Three signals from your KB analytics that tell you to restructure:
- One category gets 80% of traffic. Either the others are dead weight, or the popular category is too broad and should be split.
- Search volume is high but article views are low. Customers are searching for things they cannot find via navigation. Either the categorization is wrong or article titles do not match search queries.
- Zero-result searches reveal missing categories. If "I want to do X" appears as a search 50 times and X is not a category name, the structure does not match user vocabulary.
Helpable, Document360, and Zendesk all surface zero-result searches in their analytics dashboards. Notion does not.
How to restructure without losing SEO
Most teams worry about restructuring breaking their search rankings. Three rules keep SEO intact:
Rule 1: Keep article URLs stable. Categories can change, but article URLs (the slug at the end of the URL) should not. Helpable and most modern KB tools let you change category assignment without changing the URL slug. Confirm yours does before you start.
Rule 2: Use 301 redirects for any URL changes. If you must change article URLs, set up 301 redirects from the old URL to the new one. Search engines transfer ranking signals through 301s within 30 to 60 days. Most KB tools support 301s natively; some require manual configuration.
Rule 3: Restructure in waves, not all at once. Restructure 10 articles at a time over a month rather than 60 articles in a weekend. Smaller changes are easier to debug if traffic drops. If a wave causes ranking issues, you have a clear suspect to roll back.
What it looks like after restructuring
A well-organized 30-article knowledge base typically has:
- 5 to 6 top-level categories (verb-led, 2 to 3 words each)
- 4 to 8 articles per category (the popular ones get 8, the niche ones get 4)
- An "uncategorized" or "other" bucket avoided entirely (the existence of one means your structure is incomplete)
- Search prominent at the top of the homepage (because no structure perfectly matches all queries)
For the underlying writing rules that complement structure, see how to write knowledge base articles.
Frequently asked questions
How many top-level categories should a knowledge base have?
Three to five for under 30 articles, five to seven for under 100 articles. More than seven and customers cannot scan the list at a glance. If you have 8+ groups of content, some of them are probably sub-categories of bigger ones.
Should I use tags as well as categories?
Tags are optional and most teams over-tag. If you use them, restrict to 5 to 10 total tags across the whole KB and use them for cross-cutting concerns (audience, complexity level) not for product features. Categories should be the primary structure.
Can I have an article in two categories?
Most modern KB tools allow it (Helpable, Document360 do; Zendesk Guide does not natively). But duplicating across categories usually signals that your taxonomy is wrong. If an article truly belongs in two categories, consider whether one is a parent of the other.
What if my product changes a lot, do I have to restructure constantly?
No. Restructure when 30%+ of customer searches no longer map to your category structure, not after every product release. A good task-based structure ("Connect your data") survives feature changes; a feature-based structure ("Webhooks v2") does not.
How does Helpable's structure differ from Confluence or Notion?
Helpable is a focused publishing tool with categories, search, and a public help-center surface. Confluence is space-and-page hierarchical with deep nesting suited to internal docs. Notion is freeform with database-like flexibility. For customer-facing help centers, Helpable's flatter structure tends to work better; for internal wiki use, Confluence or Notion may fit.
Get started
Helpable's analytics dashboard surfaces zero-result searches and traffic concentration so structural problems are visible before they hurt support volume. Try Helpable free for 7 days. EU-hosted, $49 per month flat for Starter. See our SEO features for the full analytics breakdown.
Related articles
- How to Write Knowledge Base Articles
- How to Build a Knowledge Base
- Sample Knowledge Base Articles + Templates
- What Is a Knowledge Base?
- Best Knowledge Base Software for SaaS
Last updated: May 2026