We write a lot. We usually publish two articles a week on topics we know a little about, which we find interesting and which we think you enjoy too. In the last couple of months I’ve published here on LinkedIn about:
LXPs; understanding the actual workforce; the meaning, history and future of personalization; the utter indispensability of learning; AI for managers; how to visualise algorithms; the underlying kindness of recommendations; fake AI. We’ve also written on HBR about prioritizing skills, resisting algorithms and intelligent assistants.
Despite - or maybe partly because of - this, we still get asked, ‘but what is magpie?’. It happened last week: “well I've heard about magpie from your emails, but not quite sure what it is, maybe you can explain”.
It’s hard to explain, because what we’re doing is new and new stuff doesn’t, by definition, belong to an existing market category which people know and understand (and love, hate or are indifferent to). Carving out your own category is both rewarding but difficult. Related concepts are many and (in some cases) nebulous themselves:
Here are a selection of our means of describing what magpie is.
1. A recommender, like Spotify, etc.
There’s a lot of music in the world. Tens of millions of songs have been written. One of the important problems Spotify solves is to find you new music you’ve never heard but which you are likely to enjoy. Spotify uses human curation and proprietary algorithms to get good music picks to users. In this way, the issue of content proliferation in music is solved.
YouTube does the same with video, Twitter with tweets, Facebook with posts, Google with web pages, Tinder with relationships.
magpie does this for corporate learning (by understanding the learner, learning and matching the two - see below).
2. To invigorate a learning ecosystem
All clients come with baggage. Intranets, extranets, LMSs, content libraries, obsolete proprietary content, great proprietary content. Usually you - and we - need to work with all that somehow. magpie invigorates existing learning ecosystems by making intelligent recommendations at scale but for each individual, to get the right people to the relevant material.
Most learning ecosystems are not as active or impactful as their creators would like them to be.
3. As a recommendations-focused LXP
Another way of thinking about this is that we’re an LXP which is heavily focussed on recommendations. LXPs have a wide range of features and functionality. We have some of that but magpie is narrower and much more focussed on recommendations. Although we have a nice, clean user interface (UI), the focus of our work and what sets us apart is the AI, the brains behind the recommendations. It’s no guarantee but there’s a reason we successfully applied for a patent in 2014. We believe there’s some depth to the intellectual property of recommendations which is worth protecting.
At Filtered we believe the problem is really all about good recommendations and we focus all our energy on that. magpie makes good recommendations straight out of the box. The key to a good recommendation is data (to provide accuracy) and quality (to be effective). magpie incorporates large-scale external economic datasets to make personalized recommendations based on your skills/competency framework from Day 1.
The quality in magpie is powered by human curation activities that only use the best of the free content available on the web, the best of your paid-for libraries and the best of your own created, proprietary materials. So rather than create a new platform to host all your content and all your discussions, magpie simply provides a recommendation layer on top of those existing platforms which we can integrate via an API and sharing widgets into your communication tools.
We aren't planning to build a new collaboration tool because we are exclusively focussed on quality, accurate, useful recommendations. We also think it's tough to put learning in one particular platform for the workforce to get to. We prefer to take the recommendations to the places your people already go to talk and learn, be that the inbox or an existing platform (eg Slack, HR system, LMS, etc).
4. Learner, learning, match, improve
Our method is to understand the learners, the learning content, match the two and continuously improve the matching.
- Understand the learner. Pinpoint skills needs, levels and gaps.
- Understand the learning. Classify, label and structure the universe of learning content, and measure its impact.
- Match learner & learning. Match gaps to learning content with the objective of optimising productivity and best meeting learner objectives and ambitions.
- Feedback mechanism. Use explicit and implicit (usage-based) feedback from learners to improve future recommendations for them, their colleagues and their peers.
How does it actually work?
We place a layer over your existing ecosystem. That might be a chatbot widget, an iframe, a hyperlink or a box you design yourself into an interface (which we populate using our API). Users then engage in a chatbot experience so that magpie can understand them better and get them to relevant, fresh, high-quality learning, wherever they are - we don’t need or want to host anything. Through deep links and SSO, the experience is seamless.
magpie is faster to implement than a lot of other solutions, largely because we don’t need to move anything around. We can be up and running in a little over a month from a first conversation. We are ISO 27001 certified.
How does it actually help?
magpie helps learning professionals and end learners in a number of ways. Different clients get different benefits from it. Here are a few:
- Learning happens more. There’s more of a reason to engage in elective learning like soft skills when a specific recommendation has been made. magpie has achieved 50% take-up amongst its audiences - compared to an industry average of 5% for elective learning.
- Learning happens better. It’s not just more learning but better, more relevant learning as more of the right content is put to the people that need it. And it’s more engaging delivered in a conversation. Everyone likes that. Users launch learning from the top tray of recommendations 17% of the time. When they browse the rest it’s just 2%. In other words, recommended learning is 10x sticker.
- Learning professionals and the workforce generally gain firsthand experience of AI used to further their careers and have a tangibly informed opinion beyond the headlines and hype.
- Learning professionals get data insights to make better decisions. What content is working, what content can I drop, what type of content engages my people, what are my high-performers doing, where should, when in the week and in the day are people learning and can I fit my learning strategy around that. magpie enables a ‘data-driven learning design strategy’ (We’ll be publishing some findings on this later this week)
That’s what magpie is.