KB Article

Support Knowledge Base

6 min read

Also known as: Help article, Support article, Knowledge base article, Help center article

A KB article is a self-contained help document that answers one customer question or explains one product task in a searchable knowledge base.

Definition

A KB article is a structured help document published in your knowledge base to answer a specific customer question, walk through a product task, or troubleshoot a known issue. Each article covers one topic, lives at a stable URL, and is written so a customer (or an AI agent) can resolve the issue without contacting support.

Support teams use KB articles to deflect repeat tickets, onboard new users, and feed AI assistants that answer questions inside chat, email, and in-app help. A healthy KB has a clear taxonomy, consistent formatting, screenshots or short clips where useful, and metadata (tags, product area, last-reviewed date) that keeps content findable and current.

KB articles differ from internal runbooks (which are written for agents and may contain confidential steps) and from blog posts (which are marketing-led and not maintained for accuracy over time). A KB article is a living, version-controlled answer that gets updated whenever the underlying product or policy changes.

Why It Matters

Every published KB article that gets read is a ticket that didn't get filed. For a mid-market support team, a well-maintained library can deflect 30-60% of routine questions, shorten handle time on the tickets that do come in, and give your AI support agent the source material it needs to answer accurately. That directly lowers cost per contact and frees your senior agents for complex work.

When KB articles are skipped, stale, or scattered across docs, Notion pages, and old email threads, customers escalate faster, agents give inconsistent answers, and your AI assistant either hallucinates or punts every question to a human. The team ends up answering the same five questions a hundred times a week and never builds the leverage that a knowledge base is supposed to create.

Examples in Practice

A SaaS billing team writes a KB article titled 'How to update the credit card on your account' with step-by-step screenshots and a section on common errors. The article is linked from the billing failure email, surfaced by the in-app help widget, and used by the AI assistant to answer the question in chat — cutting billing-related tickets by roughly half.

A 40-person ecommerce brand maintains a KB section on returns and exchanges with one article per scenario (wrong size, damaged item, late return window). When a shopper asks about returns in the support chat, the AI agent pulls the matching article, summarizes the policy, and generates the return label without a human touching the conversation.

A B2B software company uses KB articles to support a new product launch. Before launch, the docs team publishes 25 articles covering setup, common errors, and admin settings. Sales links to them in proposals, onboarding emails reference them, and CSMs send them during implementation calls — replacing dozens of one-off Loom recordings.

Frequently Asked Questions

What is a KB article and why does it matter?

A KB article is a single help document in your knowledge base that answers one customer question or explains one task. It matters because each article that gets read or surfaced by an AI assistant is a support ticket your team doesn't have to handle manually. Over time, a strong KB compounds into one of the highest-leverage assets a support org owns.

How is a KB article different from a runbook or SOP?

A KB article is customer-facing and written in plain language for someone outside your company. A runbook or SOP is internal — it tells your agents how to handle a situation and may include confidential steps, escalation paths, or refund authority. The same topic often has both: a customer KB article and an internal SOP behind it.

When should I write a new KB article?

Write a new article when the same question shows up three or more times in tickets, when you launch a feature, when a policy changes, or when an existing article covers too many things at once. The trigger is repetition or change — not a content calendar. If your team is copy-pasting the same answer, that answer belongs in the KB.

What metrics measure KB performance?

Track ticket deflection rate, article views, search success rate (did the searcher click a result?), helpfulness ratings, and the percentage of AI assistant responses sourced from KB content. Also monitor article age — anything not reviewed in the last 6-12 months is a quality risk. The north-star metric is contacts-per-customer trending down while CSAT holds steady.

What's the typical cost of building a KB?

Cost is mostly labor. A solid KB article takes a subject-matter expert 1-3 hours including screenshots and review. A starter library of 50-100 articles is typically 100-300 hours of writing time, plus ongoing maintenance of roughly 5-10 hours per week for a mid-sized product. KB platform software itself is a minor line item compared to the writing effort.

What tools handle KB content?

Most teams use a dedicated knowledge base platform, a help-desk suite with a built-in KB module, or a docs-as-code setup tied to their product repo. The right choice depends on whether your audience is end users, developers, or both, and whether you need AI search, multilingual support, and analytics. Integrated CRM and support platforms increasingly bundle KB authoring with AI answering.

How do I implement a KB for a small team?

Start by pulling your top 20 ticket categories from the last 90 days. Write one article per category using a consistent template (problem, steps, screenshots, related links). Publish, link from your in-app help and email signatures, and review quarterly. Don't try to document everything at once — ship the top 20, measure deflection, then expand into the next tier.

What's the biggest mistake teams make with KB articles?

Letting them go stale. A KB article that contradicts the current product is worse than no article at all — it erodes trust and trains your AI assistant to give wrong answers. The second-biggest mistake is writing articles for internal jargon rather than the words customers actually search for. Always title and structure articles around the customer's question, not your feature names.

How do KB articles work with AI support agents?

AI support agents use your KB as their source of truth. When a customer asks a question, the agent retrieves the most relevant articles, synthesizes an answer, and cites the source. The quality of the AI's answers is bounded by the quality of your KB — if articles are thin, outdated, or contradictory, the AI's responses will be too.

Should KB articles be public or gated?

Most should be public. Public articles get indexed by search engines, drive organic traffic from prospects researching your product, and deflect tickets before login. Gate only content that's truly account-specific, like admin controls for enterprise plans or security configurations. A public KB is also a sales asset — prospects often read help docs before they buy.

The AMW Suite