Articles nobody opens
You put real effort into a help centre. There are dozens of articles, organised into categories, with screenshots. Your support team still answers the same questions every day, and when they send a customer a link to an article, the customer often replies with a follow-up question the article already answers.
The analytics, if anyone checks, show a handful of articles getting most of the visits and most of the rest getting almost none.
Why the help centre is ignored
| Cause | Sign |
|---|---|
| Written in your language, not theirs | Titles use internal feature names customers never say |
| Search that does not match how people ask | Searches return nothing or the wrong article |
| Out of date | Screenshots and steps that no longer match |
| Too long or too general | The answer is in paragraph six |
| In the wrong place | Customers get stuck in checkout or the app, far from the help centre |
| Gaps | The most common questions have no article at all |
The common thread is that help centres are usually written from the inside out: what the company thinks customers need to know, organised the way the company thinks about its product. Customers search from the outside in, from the problem in front of them.
What an unused help centre costs
- Support volume that should have been self-served.
- Time spent writing and maintaining articles that do not get read.
- Customers who searched, found nothing, and assumed you could not help.
- Out-of-date articles that give wrong instructions.
- An AI chatbot built on this content, which inherits every gap and error.
That last one matters more every year. Many chatbots answer from the help centre, so a weak help centre produces a weak bot.
How we turn the help centre into answers people find
- Take a sample of real support questions from email, chat and tickets, and match each one to the article that should answer it.
- Produce a gap report: questions with no article, articles that exist but did not answer the question, and articles nobody needs.
- Rewrite or create the priority articles in customers' own words, with the answer first and the detail after, and titles that match how people ask.
- Improve search, using semantic search so a question like 'can't log in on my phone' finds the article titled 'mobile authentication'.
- Bring the answers to where people get stuck: suggestions in the contact form as the customer types, help links in the checkout or app screens where questions arise, and a chat assistant that answers from the articles and links to them.
- Set up ownership and review: each article has an owner and a review date, and articles are flagged when the product or policy they describe changes.
- Track which articles resolve questions, using searches with no result and contacts that followed an article view, so the gaps keep getting fixed.
We work with the help centre platform you use, such as Zendesk Guide, Intercom Articles, HubSpot or a WordPress knowledge base, and only suggest a move if it genuinely limits what you can do.
A help centre that earns its keep
Customers find answers when they search, and more often they see the answer before they have to search at all, because it appears in the form or screen where they got stuck. Your support team links to articles with confidence that they answer the question.
Writing new articles gets easier as well. Because the gap report is built from real questions, your team knows exactly what to write next and how customers phrase it, rather than guessing which topics deserve an article.
Any chatbot built on the content gets better at the same time, because its source got better. And the gap report becomes a routine, so the help centre keeps up with the questions customers are actually asking.
Is your help centre working?
- Support answers questions that the help centre already covers.
- Most articles get very few views.
- Searches in the help centre often return nothing useful.
- Articles are out of date after product or policy changes.
- Your chatbot answers from the help centre and gets things wrong.