IF INTERESTED #194: A Framework for Building Modern Learning and Development Systems

A Framework for Building Modern Learning and Development Systems

Introduction

hi,

Corporate training platforms have repeatedly failed to deliver the necessary skills upgrades. Time and again, companies have fallen short of building the learning cultures they envisioned or executing enterprise-wide upskilling at scale. Much of this failure stems from an “output vs. outcome” mindset that persists in many HR functions, a topic I have covered in earlier post. Another contributing factor is the flawed design and technology choices behind these programs. In this post, I will propose a more structured approach to capability development and invite a conversation on how AI can strengthen learning and upskilling platforms. This was originally drafted as a chapter for my book “Blueprint for Transforming HR through AI and Evidence”, but it didn’t make the final cut among the top five sections. Instead I am covering the topic as a blog post.

Let’s start with the core problems of today’s organizations:

• Knowledge Hoarding: Expert knowledge stays trapped in individual minds or small teams

• Poor Content Quality: Most internal training content is unwatchable and ineffective

• No Personalisation: One-size-fits-all content that doesn't match learner needs or skill levels. The “personalised” content is produced by a poor algorithm and does not deliver

• Failed Communities: Platforms become either ghost towns or chaotic noise with no real value exchange

The Foundation: Four Ways Organizations Learn

Understanding how learning actually happens in organizations is crucial for building effective systems.

The well-known 70-20-10 rule, despite lacking strong scientific validation, is still a useful heuristic. It suggests that 70% of learning happens on the job, 20% through social interactions, and 10% via formal training. The exact proportions may vary, but the general principle holds.

I take a slightly different approach, organizing learning into four distinct categories. The aim is to enable more intentional management of each by first identifying them clearly.

Here are four ways we learn and develop in organizations:

1. Informative Experiences

Everything we live through in our professional life is learning. Exposure to events like a crisis to overcome in a short time, the leadership styles and behaviors we observe from our managers, the stretch assignments we are given, etc. They all develop and shape us. This kind of development is less intentional and less orchestrated.

2. Apprenticeship

Learning through guided practice under a skilled practitioner,  probably the most effective way to develop capabilities. A junior coder working side by side with senior ones on projects, a new graduate sales representative going to customer visits with senior account managers, etc. Apprenticeship is an amazing tool for development, building relationships and belonging in organizations. Unfortunately, very few organizations deliberately utilize the power of well-designed apprenticeship approaches. But that is a discussion for another day.

3. Capability Development Systems

Formal, scalable architectures that codify and distribute know-how in various formats and through different channels. For example, Udemy is a video-based capability development system. We can also include written sources like case studies, scientific papers or podcasts and in-class training sessions in those systems. These are also called formal training platforms.

4. Communities of Practice

Self-organising groups that steward knowledge and identity around a shared domain. Learning is a social act: practitioners co-create meaning, stories, and standards that no formal syllabus can capture. Demo days, online communities, pattern libraries, decision logs, and playbooks are all aspects of community-powered learning. Communities of practice are an adjacent and slightly overlapping concept to capability development systems.

Why Focus on Systems and Communities? While experiences and apprenticeship remain vastly valuable, AI and modern tools are dramatically improving the effectiveness of capability development systems and communities of practice. This represents the biggest opportunity for scalable impact in organizational learning and collaboration. 

On the other hand, better apprenticeship and a culture of creating environments of rich informative experiences are more talent management scope, that requires different solutions. Maybe a post for another time.  

The Solution Framework: The Three C's (Content, Curation and Community)

1. Content Creation

Systematically capture tacit know-how and external thought leadership in reusable formats. There is a mountain of valuable information stored in expert individuals' minds and in project teams. That information about customers, products, ways of working, and internal workflows is only shared through individual relationships or well-intentioned but unscaled events like "best practice sharing". Rarely does your true product expert have a compelling format to explain the best features of the product portfolio. Usually that information, if it is meant to be distributed at scale (like to all sales teams), is in the format of a meeting recording. An agonizing 60 minutes, where the meeting starts with background noise and the "can you see my screen" statement, and goes on with a fairly monotone lecture going through 128 slides. In that way, people who have attended the meeting will receive the necessary information through inhuman focus and determination. The people who could not attend or want to revisit the information will never be able to sit through the same painful 60 minutes of recording again. There are many experts in the organization who just avoid such exposure and share their invaluable knowledge with their closest team members only.

This is where the value of content creation comes into play. Organizations have not yet cracked a way of unearthing those potent information sources and sharing them with larger audiences to make that knowledge part of the enterprise knowledge.

Enter AI-powered video tools. The new generation of video tools like HeyGen and Descript provide an amazing array of user-friendly tools to make gripping learning videos. HeyGen can make videos from text with an ultra-realistic avatar with lip-sync and in over 150 languages simultaneously. Those tools make it very easy to edit the video, take out sentences, and replace terminology by only writing it down. They are also getting AI Agent upgrades now. A command like "please replace the first minute of introduction with background music" will be executed in the coming weeks.

No cameras, lighting, script writing, or editing agencies are needed. Just the text, and a subscription that costs 20-30 USD per month.

What this means: Every expert will have access to top-tier video creation by just writing their knowledge down.

I am not suggesting that everyone in the organization starts creating their own one-minute TikTok videos for knowledge sharing. This unlock of content creation can be governed by companies to ensure high-quality content.

Implementation Strategy:

• Guided Governance: Not everyone should create content, but the right experts should have easy access to professional tools

• Quality Standards: Implement approval processes and content guidelines

• Topic Curation: Focus on high-impact knowledge areas where expertise sharing delivers maximum value

• Template Library: Provide proven formats and structures to ensure consistency

2. Content Curation

Match the right asset to the right learner at the right moment via data-driven personalisation.

That's the fancy way of saying, can we reach YouTube-level accuracy in suggesting the next relevant video. "Relevant" here is the operative word. We need to define what "relevant" means for every employee. A few options here:

Based on their roles, interests, tenure, and skill level, the algorithm should be able to suggest the right content, and even create the content from existing libraries. For example, if someone wants to learn about decision making but they know the basics, the algorithm should suggest a highly rated video from the 15th minute onwards, because the first 15 minutes would be too basic. The algorithm will soon be able to cut and paste the relevant sections of a video based on the needs of the learner. It is pretty simple—a ChatGPT conversation with a video.

Thought leaders and curators: Organizations can also empower thought leaders or technical influencers among their employees to like and post videos, and curate playlists.

Both options above are pretty basic ideas that almost all organizations fail to put in place in an authentic way. Either the algorithms are too poor to suggest relevant content, or the platform is full of self-marketing generic content, and at the same time, real thought leaders are not selected for curation; instead, the people who are closer to HR are being promoted.

A powerful algorithm, which can be purchased and trained easily, and decent governance by the business rather than HR can make a big difference here.

AI-Powered Solutions:

• Smart Segmentation: Algorithms suggest advanced content to experienced learners (e.g., starting a decision-making video at minute 15 for someone who knows the basics)

• Dynamic Content Creation: AI cuts and combines existing content based on individual learning needs

• Thought Leader Curation: Empower internal experts to create "channels" and curate playlists like YouTube creators

• Behavioral Learning: Systems improve recommendations based on engagement patterns and feedback

3. Community

Etienne Wenger shows that Communities of Practice (CoP) accelerate learning when members share a domain, engage regularly, and produce artefacts together.

Community building has always been a topic in every company I have worked for. My first company, HayGroup, an HR consultancy company, wanted to build global expertise area communities: leadership assessment, performance management, organization design, etc. They did that with email groups and some document sharing capabilities. It was 2002 and it of course did not work. Later on, I have experienced SharePoint sites or Teams groups to build communities. Needless to say, they mostly turned out to be piles of unstructured conversations and a lot of noise. Some of those Teams groups did create good practice sharing and relationships.

Countless efforts failed because the platforms were not optimal and organizations have not governed and regulated those communities. There were no community guidelines; one post would be a valuable product update, and the next one would be celebrating a colleague's birthday. Both are relevant but probably do not belong in the same chat.

A better-suited community approach like YouTube channel owners (see thought leaders above) with people commenting on specific topics can be a better way to establish those communities.

Continuous and frequent high-value events like demo days can enhance those communities. Curation of the content again becomes the critical success factor. All similar events fail because the event content is not good enough or not engaging enough.

Even though the fundamentals that I am describing are the same as before, a significant upgrade to the platform, the change of approach to thought leaders rather than everyone having an opinion, and better governance will make communities a reality.

Modern Community Design:

• Channel-Based Structure: Like YouTube creators, thought leaders manage focused topic channels

• Clear Community Guidelines: Separate social interaction from professional knowledge sharing

• Regular High-Value Events: Demo days, expert panels, and problem-solving sessions

• Content Moderation: Dedicated community managers ensure signal-to-noise ratio remains high

• Integration with Work: Embed community participation into regular workflows

How to Start and Implement

Prep Step: Digital Decluttering (Month 0)

Recently, I have been cleaning my own storage rooms, moving for the second time in a year. Throwing out the old stuff is a sacred quest, one with utmost satisfaction.

Organizations have tons of old, irrelevant information in their digital environments. Videos of product launches from a decade ago, unconscious bias training with arguments that are now proven wrong, town hall meetings from two previous CEOs, and more. Going through those and cleaning that digital storage is a crucial step.

Step One: Discover & Design (0-3 months)

Appoint a good product manager if you haven't already. I mean a real product manager, not a learning and development specialist having a stretch assignment. I like stretch assignments, but I like expertise more. The product manager should figure out the real business need (discover phase of product management) and, based on the needs, design the roadmap for the product of "learning and development". It is better to disconnect this product team from the HR L&D team, in case the HR team wants to stick with the old paradigm of learning. Conventional wisdom should be used carefully, not fully. Protect your product team at all costs.

The most important task for the product manager is to decide on the tech stack and the governance model at this stage. The choices for the tech stack are:

• Front-End Portal: An internal "YouTube-like" progressive web app supporting video, PDF, podcast, and SCORM.

• Headless CMS: Stores metadata, version history, and peer-review status (e.g., Strapi).

• AI Recommendation Service: Pluggable API (Tealium, RTB House) that prevents vendor lock-in.

• Data Lakehouse: Unifies learning telemetry and HR data for analytics and compliance.

• Integrations: Single sign-on, Teams/Slack, HRIS, recruiting, and performance systems via xAPI events.

The governance principles and community guidelines are to be defined at this stage as well. Those decisions should ensure:

High Signal-to-Noise Ratio: Good content is key for the learning platform. The governance should be done by the business first, with a handful of content approvers that can and should veto 50% of the proposed content. In later stages, after a year or so, the thought leaders and the organic likes will promote the good content.

Community Standards: Building communities, putting them on the right platforms and keeping them alive, and shutting down the ones which did not work are all important and neglected tasks. Event management responsibilities and community content moderator roles should be appointed to people with the right common sense. These are not full-time jobs, only side assignments.

Establishing Good Thought Leaders with "YouTube Channels": Good curation should be done with the right thought leaders. Just like following people on Twitter or YouTube, or subscribing to podcasts, these thought leaders should have their own channels and promote good content based on their domain. This is not an easy task; there should be support from internal communications for those thought leaders. Who gets to be one of those thought leaders, what kind of support they need, and what a good channel playlist should look like—all of this will be designed at this stage.

External Content Platform: Connecting directly to an external vendor like Udemy can be beneficial, but those platforms can easily overtake the internal ones. The best advice is to select content from those vendors and promote that content if it is good. Which partner to use, how to promote, and how to balance that platform should be designed here. These external platforms contain a lot of good information, but they also have a lot of noise. LinkedIn Learning, for example, has provided one of the worst experiences and most irrelevant content in my previous company. Those failures should be avoided by implementing a robust filter.

External Expert Content: How to get experts from academia and the market to provide exclusive content to your company. Panel discussions and lectures from those experts are good sources of insights. What will be the monetary model, and who manages the content, these questions should be answered at this stage as well.

Executive Support: What is the role of the executives and leaders? Too much "sponsorship" makes the platform less genuine, while having no executives in any of the communities also sends the wrong signal. Rules here should be established.

The above design and governance principles are difficult to determine and implement. There is a reason why the majority of these internal learning platforms fail; putting the right amount of effort into these initiatives is crucial. Otherwise, the platform will never provide the benefit it is supposed to.

Step Two: Build & Pilot (4-9 Months)

In this phase:

• The beta portal should be live with a good user interface and experience. A/B tests and MVPs are key.

• 200+ assets should be newly produced; additional content from the clean-up should be added. Authenticity and high quality are key.

• A few pieces of good external exclusive content should be produced.

• A pilot unit and community should be selected and go live. Fast feedback loops and iterations with product management approaches, both for the tech stack, governance models, and content.

• The AI recommender engine should be in place and fine-tuned learning of preferences should start to accumulate. The data flywheel should start turning.

• Use AI to tag and categorize your content; generative AI can do this by looking at the content itself, rather than having humans tag the videos. It will boost the recommendation algorithm.

• Reward the thought leaders and community moderators to keep them engaged in the long term.

Step Three: Scale (10-18 months)

• Frictionless and elegant user experience

• Aiming for 5,000+ assets

• Company-wide go-live

• A recommendation engine that actually works (based on roles, skill needs, expertise level, and behavior). Let the AI algorithm do this; please don't start a skills definition project that will take 5 years.

• An external ecosystem of experts and academia producing a constant flow of good content.

• 85% monthly active users, and 50% weekly active users to be achieved. Minutes of use are a key metric.

This is where the delivery machine settles down and continuous improvements to the platform, algorithm, and communities are implemented.

Product management principles continue to apply. The product team should continue to work on the product going forward. There is no need to hand this over to any line organization.

Addressing Common Concerns

"This sounds expensive and complex." While the initial investment is significant, the ROI typically is several times greater. The complexity is managed through proper product management and phased implementation.

"Our employees won't create content." The key is not asking everyone to create content, but making it easy for the right experts to share knowledge in professional, engaging formats.

"We've tried learning platforms before and they failed." Most failures result from poor content, lack of personalization, and treating learning as an HR initiative rather than a business capability. This framework addresses those core issues.

"How do we measure success?" Success metrics include user engagement, knowledge transfer, performance improvement, and business outcomes. The data lakehouse architecture enables comprehensive measurement and optimization. Also, don’t get too hung up about measuring every aspect of the improvements. Investing in development is an intrinsically valid investment, that is good enough. Good CEOs understand that. 

Summary and Conclusion

Content, curation, and communities are the three C's that have been relevant from the beginning for capability-building platforms. I have never dared to implement them successfully in the companies I worked for so far. As the famous meme from Iron Man says, I was limited by the technology of my time.

Now, finally, we have access to easy and high-quality content creation through AI tools like HeyGen. We also have the ability to curate significantly better with currently available AI. If organizations combine these technological advancements with product management principles, they can establish a development platform that will boost internal knowledge, employee belonging, and collaboration. The return on this initiative is immense, if good people are put in charge.

If Interested.

Burak