Analysis of 30 help centers

SummaryHere is a short statistic about help resources of big SaaS products.

Photo of a person texting

Photo by Markus Spiske on Unsplash

If you think about launching a help center, then you might wonder how to present the content. Then this post is perfect for you.

I analyzed the help centers from top SaaS companies like Slack, Mailchimp, Zendesk or SurveyMonkey. This should give you an idea of how their help centers work.

But what qualifies as a help center? I mean there are tons of „help resources“ on product pages. To add a row to my dataset a site should be clearly identifiable as a collection of articles or pages. It should be linked from the main website. It should have a structured layout, and there should be at least 10 articles. You may be surprised at how many articles some help centers have (more on that later.)

Info: (28/30) means 28 out of 30 help centers.

The overall structure

Most of the help center content is organized into „categories” (28/30) and labeled a single resource as „article“ (26/30). Shopify however does it differently. They have a bunch of categories at first, but the content on subordinate pages is structured as a book (tree of pages.)


Starting with the intro. Most of the pages have a headline (27/30). But only 5/30 provide a sub headline. 10/30 phrase the headline as a question.

17/30 use icons to improve the look and read-ability of the pages.

9/30 show the author of the help article. If the author is public, they show the name and a profile picture.

Internationalization (i18n) is an interesting topic. If you would have asked me before the analysis, how many of the help center will have translated content. I would have answered: most of them! But the reality is that only a few translate the content into different languages (7/30).

To count the number of resources in a single help center, I used a small automated script. That is clearly not comprehensive, but necessary due to the sheer number of articles on some sites. So obviously don’t take these numbers as 100 percent correct. The average number of articles is 272. The help center with the most is Mailchimp with 788 articles. The shortest help center has 34 articles.


A “quality control” is a small feedback feature on a single article. Most of the time it is in the binary form of „was this article helpful?“ with the answers „yes“ and „no“. 24/30 track this and hopefully improve the articles based on the feedback. Some sites allow to rate a page on a spectrum (e.g. 1-10) or use emojis.

All sites included in the dataset have a search feature (30/30.) This is not unexpected, because a text search is a wonderful way to search for information.

19/30 show related entries to give some hints which article could help you too.

Tags however are a completely different story. There is only one (1/30) help center which adds small labels to the articles to improve the find-ability. Tags are short text-labels added to articles (e.g. „first steps“.) Commonly used for search results or to build listing pages to relate pages of the same topic.

18/30 provide a list of popular entries. That is a list of articles which are opened frequently and a lot of users are looking for them. Help centers show these pages most of the time on the front page.

This is the same data as a table:

# of help centers 30
„category” organization 28
„article“ label 26
Headline 27
Sub-headline 5
Question as headline 10
Icons 17
Show article author 9
I18n 7
Average # of articles 272
Quality feedback 24
Search 30
Related entries 19
Tags 1
Popular entries 18

I hope this gives you some hints on what to include in a help center. And that you get some inspiration out of it to shape your own help center 😉. There is no conclusion intentionally, just the raw numbers I collected.

If you have any comments, please send them to me on Twitter.

About the author

A blog about the evolution of Messengerturn. Announcing new features or explaining why things are done this or that way. If you like to know more, you can visit