Why knowledge search is broken, and how to fix it

An employee asks “Where is the latest NDA?” Or “How do I set up a Python virtual environment on my office desktop? Or “What’s our favorite bakery for birthdays?”

You used to know all of the answers to those questions. And it was easy for other people to ask you for answers because everyone sat together in the same small office. But now your company has swelled to two hundred, expanded across the country. Today, you wish you had a penny for every time you’ve sent out the link to the travel reimbursement form.

So you decide it’s time to create a system that helps you share the company’s evolving internal knowledge—event calendars, legal documents, IT FAQs, org charts, HR policies, etc.—so that the employees who work in those domains aren’t constantly interrupted by questions.

Adopting a traditional knowledge management solution

Here’s how initial knowledge management efforts play out at a lot of companies.

Teams that answer a lot of question from other employees—HR, IT, Product, Sales, Legal, etc.—discover that they’re answering the same questions over and over again:

Finally, a manager in one of those domains decides to do something about it. They decide to adopt a “wiki” solution like Confluence, Google Drive, or Github.

Building out the wiki is tough. Managers and their teams spend a lot of time chasing down and organizing the information. But they power through the process anyway, hopeful that once it’s over employees will be able to self-serve by finding all the answers to their questions.

Unfortunately, you ultimately find that these wiki-like solutions don’t actually solve the problem. After all of that work, you’re still getting emails and you’re still sending out links to the travel form. You wonder why?

The problem with wikis for knowledge management

Wikis fail to create effective knowledge management solutions for a number of reasons:

Here at atSpoke, we believe that interactive knowledge management is the critical missing piece.

To truly increase productivity and decrease headaches, we believe that knowledge bases must be interactive, provide answers in real time, receive feedback on the usefulness of the answers, and improve based on that feedback.

How interactive knowledge management solves these problems

An interactive knowledge management system might follow a workflow this:

In this case, user Alice asks a question via chat and gets the answer either directly from the system or from an expert Bob. The search system gets to index and train using the answer from Bob. The system also gets feedback from user Alice regarding the quality of its answers. This creates a virtuous cycle of searching, knowledge creation, training, and feedback.

Interactive systems address the issues bugging traditional knowledge management in the following ways:

When we compared atSpoke’s machine-learning-based system to a standard search system that doesn’t learn from feedback, we found that atSpoke returned the right answers at the top position in 5-15% more cases, all within the first three months. This gap will only increase over longer periods of usage.

In other areas, knowledge searching tools—like Google search—are already semi-interactive: user feedback in the form of clicks are significantly helpful in improving the quality of search results. But internal company search has not benefited from the same technologies.

atSpoke brings interactive search to knowledge management

At atSpoke, we believe interactive knowledge management is a leap forward in how knowledge is created, maintained, and discovered within companies, so we’ve been building an interactive knowledge management tool that is very easy to setup and use.

atSpoke allows users to search for knowledge without changing context, lets teams create knowledge easily and respond to questions automatically, learns from every interaction, and gets better over time at satisfying the needs of users.

Resolve tickets 5x faster with atSpoke

Let’s get started

Are you an existing atSpoke user?