View profile

Using data to measure and improve Knowledge Base success | Support Hacks

A simple, data-driven approach to creating and improving a knowledge base.
Using data to measure and improve Knowledge Base success | Support Hacks
By Sorin Alupoaie • Issue #5 • View online
A simple, data-driven approach to creating and improving a knowledge base.

Like many of you guys, I am a proud member of the Support Driven community (if you are not already, I strongly recommend you to join). Some of the recurring questions coming from Support Pros in the community on the topic of knowledge management are:

  • How do I know if my article is helping customers? Do they find the answer they need here without contacting us? How can I make it more helpful?

  • What is the next article that I need to work on? What other customer questions can we answer with a helping article? How often these questions occur?

I am also getting these type of questions from some of my own customers, so I decided to do some research and write a short guide. These are a few metrics you can use to answer the questions above.

Customer searches
Simply looking at what customers are searching for but are unable to find can be a source of inspiration. When customers are trying to find an answer and fail, it means one of these two things:
  • There is no article that answers the customer’s question. If people are looking for things you haven’t addressed, it may be worth writing an article to answer their questions.
  • The article exists but cannot be found using the search terms used by the customer. These terms can be misspellings or synonyms not included in the article. Adding these terms into the keywords section of the article will make it available for future similar searches.

Direct feedback from customers
Most probably a new article will have snippets of information missing. Make it easy for customers to ask questions and submit feedback. This will give you the insights you need to improve the documentation.
Also, look at what questions customers have when they contact your support team and mention one of the knowledge base articles. Or when they follow up from an article shared by one of your team members. They usually do this because the article didn’t help with their question or problem.
Page metrics
Tracking common web metrics for each article in your knowledge base will help you understand how good it is at helping customers. Each metric by itself won’t tell you much, but looking at all metrics can help put together a story.
Page views
Usually, you want this value to be high and trending up over time. It indicates that more customers are able to find the article and read it.
Bounce rate
A “bounce” is when someone visits a page and then leaves without taking an action, such as clicking on a link or filling out a form. The bounce rate measures how often this happens on the article page. In marketing, a high bounce rate is bad news, but in support, a high bounce rate on an article page is a good sign. It means customers found the answer after reading your article, so they stop browsing and get back to their work.
Average time on page
You want the customers to spend time on the page reading your article. Low values of this metric point to a problem.

User flow analysis
Page metrics are helpful, but understanding the entire customer’s journey as he or she navigates through the knowledge base is extremely valuable.
A user flow report offers key information on how customers are interacting with your help center by exposing the journey they take during their visit.
User flow example in Google Analytics. Credit to ricemedia.co.uk
User flow example in Google Analytics. Credit to ricemedia.co.uk
Such a report displays a graphical representation of where customers start, the pages that they visited, and where they left your help center.
Things to look out for in a user flow report:
  • High drop-off rate on the search results page.
  • Movement between high-level sections of the knowledge base without landing on a particular key article.
  • Customers landing on the contact page after visiting the help center. What articles did they read before that?
  • Searches which lead to the right article, yet still end up as support tickets.
  • Compare the user flow for different segments, like mobile and desktop users.
User flows are not usually part of the offering of knowledge base or helpdesk vendors, but they are available with third party user analytics tools such as Mixpanel or Google Analytics.

Helpdesk data
The helpdesk platform that your team uses to communicate with customers is home to valuable information. These are a few examples of qualitative data you can use to decide what to work on next in your help center.
Commonly used saved replies
If your support reps are using a saved reply frequently when responding to customers, you can turn that reply into a short article that answers the customers’ questions.
Common issues seen in support conversations
When available, the top reasons customers are contacting your support team can guide you into what the next article should cover.
Conversations with long handle times
If your support reps are spending more time than usual on certain conversations, it might be because they refer to complex topics. If you can, simplify them with a helpful article.
Articles often recommended in support answers
If you find that your team recommends certain articles frequently, ask yourself why aren’t the customers finding the article themselves. Try to picture the journey a customer has to get to the article and look out for reasons why they are not able to.

Watching for trends
Overall, these are some global metrics to watch out for.
Ratio of number of searches per new conversations created
The more customers are visiting the help center to search for answers to their questions, the less likely they will contact your support team. This ratio should increase over time as you work on the knowledge base.
Customer contact rate vs. knowledge base visits
These two values combined are often seen as representative of the overall health of a self-service system. The customer contact rate is the percentage of unique customers contacting support in a period of time out of the total number of active customers. As you gradually introduce and improve a knowledge base, this metric should decrease while the number of visits to the help center increases.

Keep At It
Success rarely comes over night. You need time and perseverance to make an impact. Although you might get a few quick wins at the start, most KB improvements will be gradual and take a significant amount of dedicated work.

I hope you enjoyed this guide and found it helpful. 😊
Sorin
Did you enjoy this issue?
Sorin Alupoaie

This is not your regular vendor newsletter. I write about new tech and proven methodologies applied to Customer Support. From Design Thinking to AI and everything in between.
An engineering twist on how to better and more efficiently serve customers.

If you don't want these updates anymore, please unsubscribe here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue