Landing Points · Practical guide
Knowledge base website: turn support into search traffic
A knowledge base website is the part of a site where a business answers the questions customers ask about its product. Split it in two and it does two jobs: the public help centre, on the main domain with FAQ schema, ranks for the questions prospects search before buying and gets quoted by AI answers, while the account-specific knowledge sits behind the login and keeps customers. This guide covers the demand behind support content in Australia, the sorting rule, five rules for making the public half rank, and how to build it.
The short answer
How a knowledge base becomes search traffic
A knowledge base website is the part of a site where a business answers the questions customers ask about its product: how it works, what it costs, how to fix it, what the terms mean. Split it in two and it does two jobs at once: the public help centre, indexed and carrying FAQ schema, ranks for the questions prospects search before buying and gets quoted by AI answers, while the account-specific knowledge sits behind the login and keeps customers. Australians search "knowledge base" about 720 times a month at $11.82 a click and "knowledge base software" 590 times at $21.11, and the traffic potential across the cluster is around 113,500 searches: support content is a search channel that most businesses have never treated as one.
Key takeaways
A knowledge base has two halves. The public help centre answers what prospects search before buying; the gated half holds what depends on the customer's own account.
Support content is search demand: "knowledge base" draws 720 Australian searches a month and the cluster's traffic potential is about 113,500. Most of it is answered on subdomains that never rank.
One question per page, the question in the title, the answer in the first paragraph, FAQ schema on the page: that is what ranks and what an AI answer quotes.
The gated half is where Gartner's 14 per cent self-service resolution moves, because it can know which plan the customer is on and show them their own steps.
The basics
What a knowledge base website actually is
A knowledge base is the organised set of answers a business keeps about its own product or service: how each feature works, what the plans include, how the integrations connect, what the security and privacy position is, what goes wrong and how to fix it, and what the words mean. A knowledge base website is one where that set of answers is a first-class part of the site rather than an afterthought in a support tool: on the main domain, one page per question, searchable, linked from the product pages, and split between a public help centre anyone can read and an account layer behind the login where the answers depend on who is asking.
The distinction from a blog or a Learning Hub is the source of the questions. A hub answers what the market asks about the subject; a knowledge base answers what customers ask about the product. The two overlap, and the overlap is where the search value sits, because a prospect comparing products asks exactly the questions an existing customer asks a support desk: does it do this, what does it cost, does it connect to that, what happens when. A knowledge base written for customers and published for search answers both at once.
One question per page
"How do I export a report" is a page, with the answer in the first paragraph. A single scrolling FAQ is one page competing for forty questions and winning none.
On the main domain
A help centre on a support vendor's subdomain lends its authority to the vendor. On your own domain it lends it to your service pages.
Split by who is asking
Public for the questions anyone might ask; behind the login for the answers that depend on the customer's plan, data or contract.
The demand
The demand behind support content
Support content is a search channel with money behind it. The figures below are from our own Snowball SEO platform, country set to Australia, pulled August 2026.
The rule
Public or behind the login
Sorting a knowledge base. The public half ranks and gets cited; the gated half keeps customers and knows who they are.
The test is the one that runs through every guide in this series: would a prospect search for this before buying? If yes, it is public, on the main domain, one page per question, with the answer written for someone who has not bought yet as well as someone who has. If no, because the answer depends on which plan the customer is on, what their data looks like, or what their contract says, it belongs behind the login, where the knowledge base can be personal: this customer's steps, this customer's export, this customer's next action. That gated half is where a knowledge base can do something a public help centre never can, which is answer the question correctly for the person asking, and it is where Gartner's finding that only 14 per cent of self-service issues fully resolve is most easily improved. The client portal guide puts this account layer in context, and the gated content guide covers why gating it costs nothing in search.
The method
Making the public half rank
Five rules. The support team already has the answers; these turn them into pages that rank.
Start from the support tickets, not the product manual
The questions customers actually ask, in the words they use, ranked by frequency. That list is the content plan, and it is the one list a competitor cannot copy. Match each question against search demand from the platform so the pages that both customers and prospects ask for get built first.
One question per URL, the question in the title
"How do I connect my calendar" as an H1 and a title of 50 to 60 characters, the answer in the first paragraph, the steps beneath it, and one intent per page. This is the same discipline as the search intent guide, applied to product questions.
FAQ schema on every page, truthfully
FAQPage structured data with the question and answer as they appear on the page, which makes the page eligible for rich results and gives AI engines a clean question-and-answer pair to quote. Never mark up questions that are not visible on the page.
Link it into the product pages and the hub
Every service or product page links to the help centre articles about it, and every article links back to the page that sells the thing it explains. A help centre that is only reachable from the footer is an orphanage; one wired into the site is an authority signal for the whole domain.
Keep it current and dated
A help article that describes a button that moved is worse than none. Show a last-updated date, review on every product release, and retire or merge articles that no longer apply. Freshness is a ranking and citation signal, and support content is the easiest content on the site to keep fresh because the support team already knows when it changed.
The next race
Knowledge bases and AI answers
When someone asks ChatGPT, Google AI Overviews or AI Mode, or Perplexity whether a product does something, the engine looks for exactly the shape a good help centre article has: a specific question, a direct answer, steps, a date, and a source that is obviously the product's own. Public knowledge base pages are among the most-cited pages on the web for product questions for that reason. The gated half, being invisible to engines, is where the answers that should not be public live, and the paywall SEO guide in this series covers the rare case where a gated article deserves an indexable preview.
There is a second, newer reason. A public knowledge base on your own domain is also the training and retrieval material for an assistant on your own site, which our agentic website guide covers. An assistant that answers from your published help centre gives the same answer the engines give, sourced from the same pages, which is the consistency across surfaces that AI visibility depends on. The $511.95 click on "ai knowledge base" is the market pricing that realisation. Our guide to getting cited by AI in Australia covers the wider discipline.
The method
How to build a knowledge base website
Export the questions
Pull twelve months of support tickets and chat transcripts. Cluster them by question. Rank by frequency. This is the list.
Week 1Sort by the rule
Public if a prospect would search it; behind the login if it depends on the account. Match the public list to search demand and build in demand order.
Week 1Build the public half on the main domain
One page per question, the question in the title, the answer first, FAQ schema, a last-updated date, links to and from the product pages. Redirect the old subdomain articles to the new pages.
Weeks 2 to 8Build the account half behind the login
The same articles, personalised by plan and data, plus onboarding and the customer's own steps. Measure resolution and return, not pageviews.
Weeks 6 to 12The order matters. The public half first, because it is the part that earns search and citations and because the support team learns the format on articles that do not depend on who is asking. The account half second, built inside the portal, with the onboarding videos beside it, so a new customer's first week and their ongoing questions live in one place.
The investment
What a knowledge base website costs
The content is the cheap part, because it already exists in the support inbox and the people who know it are on staff. The cost is the structure: the pages, the schema, the redirects from the old subdomain, the internal links, and the account layer behind the login, which sits on top of a site that starts around $2,000 to $5,000 for an Australian small business (our website cost guide has the ranges). The search mapping that decides which questions to publish first, and the monitoring that shows which help pages rank and get cited afterwards, is done by our Snowball SEO platform, which is the heavy lifting other agencies bill by the hour and the reason the budget reaches the writing and the build. See our websites and landing pages service and the content strategy service that runs the question mapping.
Sources: Keyword volumes, cost per click, difficulty and traffic potential from our Snowball SEO platform, country = Australia, pulled August 2026. Self-service resolution from the Gartner press release of 19 August 2024. FAQ structured data requirements from Google Search Central. Website cost ranges are indicative Australian market figures, not quotes. Figures were accurate at the pull date and shift over time.
Start here
Want your support answers ranking on your own domain?
Get a free audit and we will map the questions your customers already ask against what prospects search, and show you which help pages to publish first.
Get your free auditGood questions
Knowledge base website FAQs
What is a knowledge base website?
A site where the organised answers a business keeps about its own product (how features work, what plans include, how integrations connect, security and privacy, common fixes, a glossary) are a first-class part of the site: on the main domain, one page per question, searchable, linked from the product pages, and split between a public help centre anyone can read and an account layer behind the login where answers depend on who is asking.
Should a knowledge base be public or behind a login?
Both, sorted by one test: would a prospect search for it before buying? Feature explanations, pricing, integrations, security, common fixes, glossary and comparisons are public, indexed and carry FAQ schema. Steps that depend on the customer's plan, their own data and reports, onboarding, internal procedures and anything naming a client sit behind the login, where gating costs nothing in search.
Does a knowledge base help SEO?
Yes, when it is on the main domain with one question per page, the question in the title, the answer in the first paragraph, FAQ schema and links to and from the product pages. "Knowledge base" draws 720 Australian searches a month and the cluster's traffic potential is about 113,500; most of it is currently answered on support-vendor subdomains that lend their authority to the vendor rather than to the business.
Should the help centre be on a subdomain or the main domain?
The main domain. A help centre on a support vendor's subdomain builds the vendor's authority and is rarely linked from the product pages. On your own domain, wired into the service pages and the hub, every help article is an authority signal for the whole site and a page that can be cited.
How do AI engines use knowledge base content?
They cite it heavily for product questions, because a good help article has exactly the shape they look for: a specific question, a direct answer, steps, a date and an obviously first-party source. A public knowledge base on your own domain is also the retrieval material for an assistant on your own site, so both surfaces give the same answer from the same pages.
How do I start a knowledge base website?
Export twelve months of support tickets and chat transcripts, cluster them by question, rank by frequency, and sort each question by the public-or-login rule. Build the public half first, one page per question in demand order, then the account half inside the client portal beside the onboarding videos. Measure the public half by rankings and citations and the gated half by resolution and return.
Keep rolling