Customer Support & Service OperationsSetup guide4 min readUpdated September 2026

How to Set Up a Help Center: Structure, Articles and Launch Checklist

Set up a help center by starting with the questions customers already ask. Pull your most common support requests, group them into a handful of task-based categories, write a short answer-first article for each, publish them where customers will see them, and review search and ticket data every month. The tool matters less than the content and the upkeep.

Follow the six steps below in order. A first version with 20 good articles beats a sprawling one with 200 stale ones.

Vendors Covered in this Article

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

How do you decide what goes in the help center?

Don't guess what customers need. Read your last few months of support conversations and tally them by topic. Pull data from email, chat, your ticket system and, if possible, the questions your sales team hears before purchase.

Then rank topics by two measures: how often they come up and how much time each takes to answer. High-frequency, low-complexity questions are the first candidates, since they deflect the most tickets per article. Password resets, billing changes, shipping status, getting started and how a common feature works often sit near the top.

Write the list of your top 20 questions before you open any tool. That list is your launch scope. Anything that needs a human judgment call, such as a refund exception, stays out of the help center and goes to a person.

How should you structure categories?

Organize by what the customer is trying to do, not by your internal teams or product modules. "Get started", "Manage your account", "Billing and invoices", "Fix a problem" and "Integrations" are task-oriented. "Team Finance" and "Platform v2" are internal and confusing.

A few structure rules keep things findable:

  • Five to eight top-level categories. More than that and people can't scan them.
  • Two levels deep at most. Deep nesting hides articles.
  • One home per article. Link to it from other places instead of duplicating it.
  • Consistent names. If one section says "Billing" and another says "Payments", customers wonder if they're different.
  • A visible search box, since most people arrive with a specific question and skip browsing.

What does a good help article look like?

Use the same template for every article so customers know where to look:

  1. Title as the question or task, written in the customer's words: "How to change your billing email".
  2. A one- or two-sentence answer up top, so someone who only needs a reminder can stop reading.
  3. Numbered steps, one action each, naming the exact button or menu.
  4. A screenshot or short recording for anything with a complicated screen, plus alt text.
  5. What to expect after, such as "You'll get a confirmation email within a few minutes".
  6. Common problems and their fixes, in two or three lines.
  7. Related articles, to help the reader with the next likely question.
  8. How to reach a person, so nobody hits a dead end.

Write in plain language at the level of your customer. Avoid internal terms, and define any product term the first time it appears.

How do you choose and configure the tool?

Choose a tool that fits your support setup. If you already run a help desk or chat product, its built-in help center is usually the simplest choice, since articles can be suggested in the conversation and linked from replies. Intercom is aimed at teams that want articles tied closely to chat and automated answers, and Zendesk fits teams that need a more structured ticketing and multi-brand support setup. For a fuller comparison, read help desk comparison, and for B2B software specifics, help desk options for B2B SaaS.

Whichever you choose, check in a demo that you can control the design and domain, restrict some articles to logged-in customers, review analytics on searches, and export your content later. Ask about pricing and limits in writing.

How do you launch without a big splash?

Do a quiet launch first. Publish the articles, then use them yourself: for a week or two, support agents answer questions by linking the relevant article, which shows you what's unclear or missing. Fix what agents flag.

Then make the help center visible:

  • Link it in your product's help menu and in your website footer.
  • Add it to email footers and support auto-replies.
  • Suggest articles in the contact form as the customer types, if your tool allows.
  • Mention it in onboarding emails.

If you have internal procedures and customer-facing answers mixed together, decide which is which. Customer-facing articles belong in the help center. Internal how-tos belong elsewhere, and the difference is explained in process documentation vs a knowledge base.

How do you keep it accurate and prove it works?

Give every article an owner and a review date, and update articles when the product changes. Add a habit: when an agent answers the same question twice, they either link an article or write one.

Each month, review a short set of numbers:

  • Searches with no results. These are topics you haven't covered or words you haven't used.
  • Most viewed and least helpful articles, using any helpfulness rating your tool provides.
  • Tickets on topics that have an article. If people still write in, the article is hard to find or doesn't answer the question.
  • Contact rate per customer over time, which should fall as coverage improves.

For teams exploring automated answers built on top of this content, the guide on internal knowledge base agentification covers the preparation, and Zendesk vs Intercom for IT consulting shows a services-firm view. Whatever you add later, a clean, current set of articles is the foundation.

Executive Capability Standard

What Good Looks Like

A customer help center built from real ticket data, with task-based categories, answer-first articles, named owners and a monthly review of searches and deflection.

Building The Capability (5-Stage Skill Ladder)

1. Learn:Read three months of support conversations and tally the questions by topic and time spent.
2. Do Manually:Write your top 20 articles from a single template, and have a newcomer test each one.
3. Delegate:Assign each article an owner and review date, and make agents link an article whenever they answer a repeat question.
4. Automate:Suggest articles inside chat and contact forms so customers see answers before they submit a ticket.
5. Buy:Choose a help desk or chat platform with a built-in help center once your ticket volume justifies a dedicated tool.

How to Get Started

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

Intercom

Fits when you want help center articles tied to chat and automated answers.

Visit Intercom→
Zendesk

Fits when you need a structured ticketing system with a help center, possibly across several brands.

Visit Zendesk→

Frequently Asked Questions

How many articles does a help center need at launch?

Twenty to thirty well-written articles covering your most common questions is enough to start. Choose them from real support data, not guesswork. Add more as new questions arise, and retire articles that no longer apply. Quality and accuracy matter more than volume.

What is the difference between a help center and a knowledge base?

A help center is customer-facing and answers questions about your product or service. A knowledge base can be internal, holding procedures and reference material for staff, or external. Many companies run both, and keeping them separate avoids exposing internal details to customers.

How do I know if my help center is working?

Watch the volume of tickets on topics that have articles, searches with no results, article ratings and the overall contact rate per customer. If tickets on covered topics fall and customers find answers without writing in, it's working. If not, improve titles, search and clarity.

Should we use AI to write help center articles?

It can speed up drafts, but a person who knows the product must check every article for accuracy before publishing. Wrong instructions cost you tickets and trust. Use AI for first drafts and rewording, and keep an owner and review date on each article.

About the numbers

This guide doesn't quote a sourced benchmark. Figures in it are estimates or general guidance, so check them against your own numbers.

Related Guides