Learning Insights | Filtered | Content Intelligence

Finding things, part II

Written by Toby Harris | Jun 28, 2024 8:31:51 AM

A few weeks ago Filtered’s CEO wrote about the high cost of finding things in personal life for his blog, One Thing At A Time. He estimated that we might gain as much as four minutes back a day just by knowing where things are.

Various sources over the years have suggested that finding things - digital information rather than physical objects - occupies a far larger part of our working day. Before we get carried away with this, the oft-cited figure that knowledge workers spend 2.5 hours per day frustratedly searching is a serious misinterpretation. But as the author of that article points out, the better and connected question is whether the time spent searching is ‘time well spent’ compared to other tasks.

And time spent searching in an LMS or HR system? This is rarely time well spent. 

Don’t get me wrong. Talent development functions have bigger problems: their abiding tie to discredited forms of compliance training and lack of business relevance are two big ones. But it is undoubtedly the case (and certainly has been for the 15 years or so that I have worked in this area) that time spent on painful, fruitless searches is the major technology hurdle for L&D.

Search is a problem on which fresh problems arise. People can’t find stuff, so we curate content into academies and intranet sites only for people to get lost in those. People can’t find stuff, so we implement large scale employee experience platforms, and people get lost in those too. 

As the consumer experience of search grows better and better, and whole platforms spring up on the basis of discovery (AirBnB, Booking.com, Amazon), the frustration with poor search at work grows. Poor search creeps up on enterprise learning platforms: they are rarely designed to start from a good search experience and, crucially, the quality of the metadata going in is poor.

Bad time spent searching is a concrete, immediate symptom of content chaos. Your internal customers are likely unaware of the millions you have wasted on their behalf on content they won’t ever use (phew!). But when these people search for ‘remote teams’ and they receive a set of results about Microsoft Teams, or when they have to wade through a dozen similar sounding project management courses, they feel the $9bn cost of content chaos on a personal level. 

In building our content intelligence technology, we set out to assign relevance scores to any piece of content based on any kind of topic or skill definition. In doing so, we ended up creating the finest search engine in the L&D industry. It is a fine search engine because: 

  • It can understand any kind of natural language query, spelling mistakes included
  • It’s now super fast, with coding that allows it to parse 300,000+ assets in real time
  • Uniquely, the relevance scores gives you a sense of confidence in the results


Up until now our industry-leading search has been buried in different functions: the curation page, for instance, or the skill configuration page. No longer. Search is now central to the user experience. Every user, whether they have our gen-AI features turned on or not, will get real time, contextual, natural language search with relevance scores. Even better, the gen-AI-enabled users get a coach and guide configured to help them find exactly what they really need rather than what they think they need.

Better still the search is available as an API endpoint: ie, another platform sends us a query, and we hurry back with relevant results (and relevance scores) in microseconds.

If you’re a Filtered customer the search page is live in your instance now. It’ll be used to sift through the thousands of validated content items. And if you want to rebuild your learner experience around excellent discovery, talk to us.

Finding things at work is no longer an unsolvable problem.