How Knowledge Bases and Search
Have Failed Customer Service

Many service organizations rely on knowledge base articles to troubleshoot product issues, especially if they support highly technical or complex products with various error codes, flags, and malfunctions.

A call center agent may identify a common issue as a good candidate for a knowledge base article or a workflow within a CRM might surface a case. If a simple issue keeps repeating, it seems like a good idea to publish an article about it. Or, if the resolution to a complex issue involves a certain function, caveat, edge case, or error code, capturing the detail seems useful to reduce the time and cost to fix when the issue recurs.

A knowledge base may contain hundreds, or thousands, of articles that are supposed to help customers, agents, and technicians figure out why a product isn’t doing what they expected it to do – and how to fix it.

But there are a few problems:

Long Workflows Ensure Stale Content

The process from seeing the need for an article to getting it in the production queue, drafted, reviewed, approved, and published can take 2-3 months or longer.  So the information in the article may be stale before anyone even reads it. Meanwhile, the issue is still lingering and customers are still encountering it and support teams are still reporting it and the contact center is bogged down with calls. It all starts to pile up because of that lag. 

Too Much Knowledge (Not Much Wisdom)

The other issue with the current state of knowledge base development is there’s simply too many knowledge base articles. Call center agents or field service technicians have to dig to find the right article for the current ticket they’re working–and quickly so as to not keep the customer waiting too long. 

Then parts of the article become irrelevant as an issue is remedied or circumstances change, so articles have to go through a whole new update cycle. New articles may be created in the meantime and older articles may not be edited or removed, adding to the mess.This is how knowledge management becomes inefficient.

Knowledge Base Search Doesn’t Work

And perched on top of all this mess, you have a search box. The underlying search application for many enterprise systems leaves a lot to be desired. Most of these search apps can’t do much beyond rudimentary keyword-based search. 

For example, when searching for “What is the latency for Core 1000?” the system may find “core 1000” in an article but not understand that those two words are a product name. It’ll return articles with the word “core” and articles with the number “1000”, but isn’t smart enough to understand the intent of the query.

In most organizations, 25% to 30% of incoming queries from customers are “how to” questions (and this percentage increases in more technical environments).

To resolve an issue, service teams may need a how-to article, or they might be looking for a very specific data point, or maybe something buried deep in the middle of the product manual that’s in a table or diagram. 

Some search apps can’t even screen out common words in queries like “how do I” and return any result where the word “how” is used. For the most part, search is not up to the task of finding answers in a knowledge base. 

Compounding Service Inefficiencies–and AI as an Answer

When you combine these issues: product complexity, a lag in knowledge base article production, stale content, and poor search, the result is a continually decaying knowledge base that is somewhere between inefficient and unusable. Customers, agents, and technicians are left struggling to find answers, sifting through thousands of documents that are out of date and impossible to search.   

To solve this problem, service organizations need a new way to manage knowledge, and a new way to access it. Neuron7 solves these problems, using AI to analyze data from any source – product manuals, case notes, knowledge base articles – to create a “resolution system of record” that continually learns and is easily searched. Neuron7 provides a single source of truth that can be accessed anywhere, by anyone to diagnose and resolve issues in seconds. 

If you’d like to learn more, contact info@neuron7.ai to talk with an AI expert and see a demo.

Contact Us

Resolve issues faster with AI that understands your service data and continually learns as you use it.