Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

SEO › SEO Content & Strategy

Help Centre and Documentation SEO

How help centres and product documentation earn search traffic: where they live, writing articles people search for, versioned docs, private content and links.
Guideintermediate10 min readUpdated

Why support content is search content

The support channels and self-service guide treats the help centre as a support channel and notes that customers find articles three ways: the site's own search, links at the point of need, and search engines. This guide is about the third, which is often the largest and the least managed.

Support queries have a recognisable shape: the product's name with a task ("export invoices to a spreadsheet"), an error message copied word for word, a feature with "not working", or a question about a limit or a setting. The Search Intent is unusually clear: the searcher wants an answer now, and the best page is the one that gives it fastest. When a vendor's own article does not rank, the searcher finds a forum thread, a third-party tutorial written against an old version, or a competitor's article explaining how their product does it instead.

Support content also serves people who are not customers yet. Technical evaluators read documentation to check whether a product can do what they need, and a clear, current article answers a sales question without a call. That is why help content belongs in the site's search planning, not only in the support team's backlog.

Google's helpful content guidance describes content created for people first, which demonstrates expertise and leaves readers feeling they have learned enough to achieve their goal. Good help articles meet that description by their nature. The search work is mostly about making sure they can be found, are not undermined by duplicates, and are written in the words customers use.

Where the help centre lives

Help centres often run on a vendor's platform, and the platform's default address decides where they live unless someone decides otherwise. There are three options:

LocationStrengthsWatch for
Subfolder of the main domainShares navigation, links and standing with the main site; one property to manageNeeds a reverse proxy or a platform that supports it; more set-up work
Subdomain of the main domainEasy to attach a vendor platform; separate release cycleLinks between help and the main site have to be built deliberately
The vendor's own domainNo set-upVisits, links and standing build up on someone else's domain; avoid

The subfolder versus subdomain comparison sets out the trade-offs between the first two. Whatever the choice, check the basics that help platforms often get wrong: public articles readable without logging in, content present in the HTML rather than loaded only after interaction, one host name serving each article (not the vendor's address and yours in parallel), a sitemap listing the articles, and language versions handled consistently. Moving the help centre later is a site migration, with every article URL needing a redirect, so the decision is worth getting right early.

Writing articles people search for

Google's SEO Starter Guide suggests thinking about the words people might use to search for your content, and notes that people who know a topic well use different words from people new to it. Help content lives in that gap. The product team names a feature one way; customers describe what they are trying to do in another, and the article has to meet them there.

  • Take the wording from customers. Support tickets, chat transcripts, the help centre's own search log and Search Console queries all show how customers phrase problems. The questions asked most often, in the words used most often, are the articles to write first.
  • One question per article. An article titled with the task or question it answers, so the title shown in results tells the searcher they have found it.
  • Answer first. The answer or the first step in the opening lines, with background afterwards. The writing for the web guide covers front-loading, headings, plain language and scannable steps.
  • Error messages as text. Quote the exact message in the article body, not only in a screenshot, because the searcher is pasting that text.
  • Say what it applies to. The product, plan, version and platform the article covers, and when it was last checked.
  • Steps that match the product today. Screenshots and labels that match the current interface; out-of-date steps generate contacts rather than preventing them.
  • One article per question across products. Where several products or editions share a task, write one article with a section per difference rather than near-identical copies.

The same habits serve an AI assistant that answers from the help centre, which reads articles as separate passages; the tutorial on preparing documents for an AI assistant adds the steps that matter there, such as one current version per fact and tables explained in sentences.

Versioned and product-specific documentation

Software documentation multiplies. Each release can produce a new copy of every page, and after a few versions a search for a task may return the documentation for a version two releases old, with steps that no longer work.

Google's documentation on consolidating duplicate URLs explains how to indicate which of several duplicate or very similar pages is preferred, using a Canonical Tag, redirects or sitemaps, and that its systems treat these as signals when choosing the URL to show. It also advises against using noindex to choose a canonical within a site, because noindex removes the page from search entirely; a canonical is the tool for that job. Applied to versioned documentation:

SituationTreatment
Current version of a topicA stable URL that always holds the latest content, in the sitemap, linked from navigation
Older version substantially the same as currentCanonical to the current page, with a banner linking to it
Older version that genuinely differs and customers still useIndexable, labelled with its version in the title and heading, linked to the current page
Version no longer supportedRedirect to the current page, or keep as an archive marked unsupported
Same article for several editions or plansOne article with sections per edition, rather than copies

Two habits keep this working. Internal links, navigation and the sitemap point at the current version, so the signals the site sends all favour it. And the version pattern is decided once, in the documentation platform's configuration, rather than page by page when someone notices old docs ranking.

Logged-in, private and community content

Not all support content should be public, and not all public content should be in search results.

  • Account-specific and confidential content. Content about a customer's own account, internal procedures for support agents, and documentation for unreleased features belong behind a login or other access control. Google's noindex documentation describes noindex as a way to keep a page out of search results; it is not access control, and anyone with the address can still read the page.
  • General how-to behind a login. A common mistake is to put the whole help centre behind sign-in, which hides every article from search and from evaluators. Keep general product help public.
  • Public pages that do not belong in results. Thin category listings, printer-friendly copies and search result pages within the help centre can carry noindex. Google's documentation notes that the page must not be blocked by robots.txt, or the crawler never sees the rule.
  • Community forums. Customer forums answer the long tail of questions no article covers, and good threads rank. They also accumulate unanswered questions, duplicates and spam. Moderate, mark accepted answers, merge duplicate threads, qualify links users post (the outbound links guide covers the ugc value), and turn questions that recur into proper articles that the threads then link to.

Linking help content to the rest of the site

Help centres are often islands: a separate platform, a separate team, and a single link in the footer. Crawlers and visitors reach articles mainly through the links pointing at them, so the island stays poorly connected.

  • From the main site to help. Feature pages link to the documentation for that feature; pricing links to plan and billing articles; the navigation or footer links to the help centre's main categories, not just its home page.
  • From help to the main site. Articles link to the feature page for the capability they describe and, where relevant, to plans that include it, so evaluators reading documentation can find the commercial page.
  • Within help. Category pages act as hubs listing their articles, and related articles link to each other with descriptive anchors. The topic clusters guide model fits: the category overview as the hub, articles as the supporting pages.
  • From the product. Contextual help links in the interface do not build search visibility directly, but they make customers use and trust the articles the search work is improving. Support replies are another route in: the tutorial on building a library of saved replies links each reply to the article that holds the detail rather than copying it. Help videos follow the same routes, and the guide to building a customer help video library links each video from the matching article and the product screen where the question arises.

Every Internal Link should use the article's current, canonical address, not a redirect or an old version. For software companies, the SaaS SEO guide places documentation among the other page types a software site needs.

Help centre search is measured by whether customers find answers and stop needing to ask, not only by visits.

QuestionDataSource
Which questions bring searchers to help content?Queries, impressions, clicks and position for the help sectionSearch Console, filtered to the help path or property
Which articles are found but not chosen?Articles with impressions and few clicksSearch Console; titles to rewrite
Which questions have no article?Ticket topics and search terms with no matching article, compared with queriesSupport system, help centre search log, Search Console
Do search visitors get their answer?Contacts raised after visiting from search; article ratingsAnalytics joined with the support system
Are old versions taking the visits?Clicks to older version URLs compared with current onesSearch Console by page

The support channels guide measures the help centre by what it prevents and what it fails to answer; search adds a large group of customers who never use the help centre's own search box. Read both together. To judge help content by whether its readers stay rather than by visits, the guide to content for customer retention compares churn among readers with matched non-readers.

Illustrative example — not a real result

A scheduling software company finds its old docs outranking the new

Sparkthimble Scheduling, an invented appointment-scheduling software company, keeps its documentation on a subdomain with a copy of every article for each of its last four major versions: 4 versions of 350 articles, 1,400 indexed pages. Search Console shows that for the 50 most-searched tasks, a version-two or version-three article ranks above the current one in 31 cases, and support records 120 tickets a month from customers following outdated steps. The help centre's only link from the main site is in the footer. The team creates stable "latest" URLs for every article, redirects all 350 version-one pages to them, adds canonicals from the 540 version-two and version-three pages that closely match the current text, keeps 160 genuinely different version-three articles indexable with version labels, and links each feature page to its documentation. Three months later, 520 documentation pages are indexed, a current article ranks first for 46 of the 50 tasks, visits to help content from search rise from 18,000 to 26,000 a month, and tickets citing outdated steps fall to 30 a month.

Common mistakes

  • Leaving the help centre on the vendor's domain. Every visit and link builds someone else's site.
  • Putting all help behind a login. Customers and evaluators searching for answers never see it.
  • Writing in the product team's vocabulary. Customers search in their own words and with exact error text.
  • A copy of every article for every version. Old versions compete with current ones and send customers to steps that no longer work.
  • Using noindex to hide confidential pages. Noindex is not access control.
  • An island help centre. One footer link from the main site and none back.
  • Measuring page views. Contacts prevented and questions left unanswered show whether help content works.

Frequently asked questions

Should a help centre be on a subdomain or a subfolder?

Either can work. A subfolder shares the main site's navigation and standing more easily, which suits help content that also supports evaluation and sales. A subdomain is often simpler when the help centre runs on a vendor's platform. The subfolder versus subdomain comparison sets out the trade-offs. What matters most is that the help centre sits on your own domain rather than the vendor's, is crawlable without logging in, and is linked from the main site.

How should old versions of documentation be handled for search?

Keep one current page for each topic at a stable URL, and decide for each older version whether it is substantially the same or genuinely different. Near-identical older versions can point to the current page with a canonical tag, which Google's duplicate URL documentation describes as a way to indicate the preferred version. Versions that customers still need and that genuinely differ can stay indexable, clearly labelled with the version and linked to the current page.

Can private or internal help articles be kept out of search with noindex?

Noindex keeps a page out of search results, but it does not stop anyone who has the address from reading it. Anything confidential, such as internal procedures, unreleased features or account-specific content, belongs behind a login or other access control. Use noindex for public pages that are useful to people already on the site but should not appear in results, and remember that Google can only see the rule if it is allowed to crawl the page.

Terms used on this page

Frameworks

Where this page is used

Sources

  1. SEO Starter Guide — Google Search Central
  2. Consolidate duplicate URLs — Google Search Central
  3. Block Search indexing with noindex — Google Search Central
  4. Creating helpful, reliable, people-first content — Google Search Central

Related