Organizations have more content than they know what to do with. It’s become incredibly easy to create and store documents—and endless versions of documents—as well as wiki pages, meeting recordings, videos, data files, employee profiles, and other knowledge artifacts. While this mountain of information has upside potential, it can cause a host of problems, from expensive storage fees to ramifications for legal discovery if the organization is investigated or sued. But the biggest problem is for employees themselves, who find it progressively harder to find what they need.
Software vendors and organizations are working hard to solve this problem. Solutions range from AI-fueled search functions to chatbots that retrieve answers for employees and then refine those answers through iterative conversations. But one of the most exciting developments is the advent of smart recommendation systems, which suggest relevant content and colleagues to users based on context.
The Rise of Smart Recommendation Systems
Smart recommendation systems integrate information about users and their past behavior to filter items and predict which will be most relevant or appealing to a particular user. The systems may be applied to push relevant items to the user in the course of their work or to drive the most relevant items to the top of search results. These systems are considered “smart” because they include embedded machine learning algorithms that allow them to improve based on user feedback and adapt to new user behaviors.
Smart recommendation systems are technically not new—APQC has studied KM leaders like Wipro and MITRE that had them in place five or six years ago. But cloud platforms, along with advances in machine learning and AI, are making them a more affordable mainstream option.
In APQC’s 2019 Enterprise Content and Collaboration in the Cloud survey, we asked respondents which emerging technologies and capabilities motivated their organizations to move to cloud platforms like Microsoft Office 365 and Google G Suite. Smart recommendation systems for both content and colleagues were at the top of the list, with 53 percent and 40 percent of respondents citing them as driving factors, respectively. I’ve selected this as APQC’s October KM Metric of the Month to highlight the growing interest in these tools and explore their implications for knowledge management.
Why Smart Recommendations Matter for KM
Smart recommendation systems can improve enterprise search and discovery by ensuring that, even as the volume of content continues to grow, the top results are relevant and tailored to each user. In organizations where “broken search” is an established pain point, this can solidify KM’s business case and help it demonstrate its value in the form of time savings and reduced frustration.
But perhaps more importantly, smart recommendations help transition employees from a search-based paradigm (where they have to seek out every piece of information) to a more expansive concept of “knowledge discovery.” Recommendation systems proactively deliver content to users based on who they are, what they’re working on, and their prior interactions with the corporate intranet, work applications, and KM platforms (e.g., knowledge or colleagues they’ve searched for or interacted with). When these recommendations work right, employees don’t have to step away from what they’re doing to look for knowledge or expertise. Instead, relevant items just pop up in the systems they are already working in—ideally, at the exact moment of need.
All this integrates KM into the flow of work and can encourage the uptake of available knowledge across the business. If KM engagement increases and the organization is better equipped to surface and use the knowledge it has collected, then KM’s star rises and its raison d’être becomes more self-explanatory.
What Challenges Lie Ahead?
Of course, smart recommendation systems aren’t perfect—especially those commercially available right now—and there are caveats to consider.
For example, these systems often require integration across multiple platforms, including ones outside KM’s control (such as HR systems and customer data systems). As a result, technology integration, security, and privacy concerns can arise. In addition, some organizations may not have enough data to obtain accurate recommendations, especially when dealing with technical content that is used by a small subset of the workforce.
These challenges are amplified in off-the-shelf solutions like Microsoft Delve, which is part of the Office 365 platform. Admittedly, some organizations are happy with what Delve offers. But others have piloted it only to find that its suggestions for relevant content, projects, and colleagues to connect with are not sufficiently fine-tuned to be useful. And bad recommendations that create noise in the system are worse than none at all.
My advice is for organizations to proceed with caution and perform their own pilots to see what works. The need is clear, and the technology is catching up quickly. A lot of organizations are investing in smart recommendation systems, and KM leaders need a seat at the table as these decisions are made. KM can make sure the systems are implemented thoughtfully, with existing knowledge repositories and established user requirements in mind. Moreover, it’s vital for KM teams to insert themselves into projects with the potential to move search and discovery capabilities forward—otherwise, they may find themselves and the systems they manage becoming less relevant to organizational knowledge flow.
For more information on this topic, check out Smart Recommendation Systems: Emerging Technologies for Knowledge Management, part of the Emerging Technologies for Knowledge Management collection.