Knowledge Base Article Count Recommendations - Finding the Right Balance

[Kohei Tominaga1]

Hi ServiceNow Community,

I'm leading a knowledge management workshop and we've encountered a common question regarding the optimal number of knowledge base articles. The concern raised is that a large number of articles could negatively impact searchability. Users might encounter numerous articles addressing similar issues, leading to confusion and wasted time sifting through irrelevant results. This is especially problematic when search terms are not highly specific.

My current thinking is that there shouldn't be an arbitrary upper limit on the number of knowledge base articles. Instead of focusing on a specific number, we should strive for comprehensive coverage of all relevant topics, avoiding redundant articles as much as possible. We should aim to provide answers to as many questions as possible, ensuring that each article addresses a unique problem or situation.

To address the searchability concerns, I believe we should focus on:

• Leveraging analytics: Closely monitoring metrics like click rank and utilizing AI Search dashboards to identify articles that are underperforming or frequently confused with others.
• Continuous improvement: Regularly reviewing and refining articles, consolidating similar content where appropriate, and adding or improving metadata (tags, categories) to enhance search accuracy.
• Improving search functionality: Tuning AI Search Functionality to enhance the search capabilities of the knowledge base by adding synonims

I'd love to hear your thoughts and experiences on this topic.

[Dhruv Chandan]

Hi,

You concern about optimizing knowledge articles is quite common during greenfield implementation. However, the effectiveness depends more on how the articles are structured and maintained rather than the actual number of articles.

For example:

1. Metadata: Enhancing metadata by adding accurate tags, categories, and structured data improves both search accuracy and content discoverability.
2. Feedback: Implementing a feedback mechanism for each article allows the BAU team to track and maintain relevance over time.

3. Ownership: Assigning managers for each knowledge base ensures their teams remain accountable for ongoing maintenance.

There can be definitely other factors that can be discussed however they can be tackled as the discovery workshop progresses with the client.

Thanks,

Dhruv

[David Kay]

Your response to the "too many articles" is 100% right.  How often do you hear people say "Google indexes too many web pages?"  Never--all we care is if useful information is on the first page of results, ideally above the fold.  How many articles should you have?  Enough, but no more.

Findability is what matters.  Are articles clear, crisp, and about one thing?  Have we captured the words and context of the person who needs this article, especially in self-service?  As @Dhruv Chandan says, is our classification of the article correct?  Are we avoiding dupes and fixing them when we find them? If so, and if as you say you're keeping up with search tuning, there can't be too many articles.

One additional thing to do is to automatically archive articles that aren't being used, where "aren't being used" is a specific number of self-service page views and incident / case attaches over some longish period of time--those criteria are pretty specific to your cycles of adoption and volume of use of the KB.  That gets rid of stuff that is likely not relevant, although you can always republish it if needed.

Anyhow, you're right, and if you would like me to put that on our letterhead to show your execs, I'd be happy to.  🤣

[Eoghan Sinnott]

Hi,
I would agree with the points made above. There should be no upper limit of articles, you just need to make sure that you have the right articles in place at the right time. The KCS methodology of Knowledge Management preaches the approach of having knowledge 'just-in-time' instead of 'just-in-case'. You don't want to clog up your knowledge base with issues you THINK someone might have, you need to make sure that you have the information that people are looking for and in the clients context. Using the various performance analytics and dashboards that are available to monitor usage, search terms, click ranks and feedback will help ensure you have a healthy knowledge base
As @David Kay mentioned, removing articles that aren't being viewed, linked to incidents/cases, or aren't receiving any feedback is a practice that you need to implement.

When it comes to the AI search functionality, the synonym dictionary is one that we have found to be of great advantage. Instead of having to load up individual articles with specific Meta terms, you can utilize this to cover similar terms across multiple articles.
You should also take a look at the Result Improvement Rules if you need to tailor searches for specific key words or queries.