You can use InfraNodus to analyze the whole content of your Zendesk knowledge base. It can be very useful for improving the content of your help center and for search engine optimization.
For instance, you may discover that some of the topics that you consider important are not covered in your help center articles. You also can use the built-in AI to come up with new interesting ideas for your support portal content.
Below we will demonstrate how you can do that using our own knowledge base as an example.
Step 1: Export Your Zendesk Knowledge Base
Zendesk doesn't provide an easy-to-use export script for their knowledge base. However, you can use one of the apps from their marketplace or their API to export the data.
Here is a link to the Python script we developed for exporting your knowledge base data from Zendesk. It can also be useful to run it to back up your data:
Just follow the instructions in the README file, choose the MD export option, and you will have a nice and clean folder with all your knowledge base articles in .MD format, without the HTML formatting.
Step 2: Import the Data to InfraNodus
Use the MD import app (e.g. Obsidian / Roam Research / LogSeq) and select all the knowledge base articles that you exported from Zendesk.
Step 3: Visualize the Results as a Graph
After a few seconds, you will see a graph that will visualize both the content of your knowledge base and the page titles:
While you can, use InfraNodus to analyze the links between your knowledge base articles and to improve the connections and cross-references between them, this is outside of this article's scope.
Instead, we will only focus on the content of your articles in order to find out what they are about and what may be missing.
Step 4: Text Analysis of Your Zendesk Knowledge Base
Switch to the Concepts Only view as shown above and you will have a graph of the concepts used in your knowledge base:
On the right-hand side in the Analytics panel you will see the main topics present in the knowledge base.
1. graph node add
2. infranodus import setting
3. text network discourse
4. idea topic gap
In general, this knowledge base is about
1. adding the nodes to the graph,
2. import functions,
3. text network and discourse analysis, and
4. generating new ideas using the gaps between the different topics.
We can also see that the graph's structure is biased, mainly because it is too focused on 1 word that is used a lot: "graph".
Actionable insight #1: use the word "graph" less, in order to improve search engine ranking for other terms
Step 5: Zendesk SEO Optimization: Find What's Missing
Let's remove the word "graph" from the network to see the topics hiding underneath:
The results above are pretty good, however, we can also see that we're missing quite a few important topics that we would like to attract the users' attention to:
1. text analysis
2. text mining
While the word "text" is the most influential now in the graph, the word "mining" is not shown in the graph and "analysis" is not very influential as well: if we search for it in the search box at the top, we see that it's not so relevant —
Actionable insight #2: add more content that talks about "text analysis" and "text mining" to the knowledge base.
Step 6: Find the Gaps in Your Zendesk KB Using the AI
You can also use the built-in GPT3 AI and network structure insights to find the gaps in your current knowledge base discourse. InfraNodus will identify the topics that could be better connected and propose to bridge the gaps between them using AI-generated questions or facts. You can use them to develop new content ideas for your support portal.
For example, in our case:
Go to the "Gap Insight", then click "Generate a Question" under the structural gap identified by InfraNodus.
After a few suggestions, you might see something interesting, just like in this case:
"How can we measure the influence of certain nodes in order to generate insights about the main ideas?"
Actionable insight #3: add more help articles on using the individual nodes' measures to generate insights about the whole discourse
If you're interested to try out this approach, go to https://infranodus.com and set up an account there. You can then export your knowledge base data and import it into InfraNodus to perform advanced text analysis of your Zendesk data.