Introduction
Adaptive learning platforms promise to tailor content and pace to each individual, offering a personalised path through material. Whilst the concept holds great appeal, true adaptivity relies on rich data: not just quiz scores and completion rates, but metacognitive indicators such as time-on-task, problem-solving approaches, and learner confidence. Many organisations, however, remain stymied by legacy HRIS systems and siloed platforms that cannot communicate effectively, severely limiting the potential of AI-powered adaptivity. This article explores how L&D leaders can push beyond these constraints and harness the full spectrum of learner behaviour data. More importantly, data management and hygiene, as well as data literacy, are key readiness and success factors to ensure AI implementation more broadly, and are part of the overall management of a modern organisation. The key question to ask yourself is: what does it take to run a modern organisation in 2025, and are your data capabilities up to the task?
What is Adaptive Learning?
Adaptive learning is an educational approach that uses technology to dynamically adjust learning content, pace, and difficulty based on a learner’s performance, behaviour, and preferences in real-time. Unlike traditional one-size-fits-all learning methods, adaptive platforms create personalised learning pathways through sophisticated analysis of learner interactions and responses.
The system works using artificial intelligence and machine learning algorithms to understand each learner’s needs, understanding, strengths, and challenges. As learners interact with the platform, it builds a comprehensive profile of their learning style, pace, and preferences, using this information to optimise the learning experience.
At its core, adaptive learning relies on several key components working in harmony. Content sequencing ensures intelligent ordering of learning materials based on mastery and prerequisites. Dynamic assessment provides continuous evaluation that adjusts question difficulty based on responses. Personalised feedback delivers targeted suggestions and explanations tailored to individual understanding. Finally, comprehensive learning analytics offer detailed insights into learning patterns and progress, allowing for continuous refinement of the adaptive process.
Through this sophisticated combination of technology and pedagogical principles, adaptive learning platforms can create truly personalised learning experiences that evolve with each learner’s progress and needs, making education more efficient and effective for everyone involved.
Behavioural Metacognition
1. Beyond Correct/Incorrect
Traditional e-learning tests only whether an answer is correct. Adaptive platforms incorporate deeper metrics, i.e. how long a learner deliberates on a question, the sequence of attempts, and the types of hints or feedback they request (Bransford, 2020). This yields insight into not just what learners know, but how they think about problems, revealing metacognitive traits like self-awareness and strategic thinking.
2. Real-Time Adjustments
With these richer data streams, AI can adjust the difficulty or provide targeted remediation immediately. For example, if a learner consistently hesitates on advanced finance concepts, the system might serve a quick refresher on the basics before moving on. Over time, the AI develops a profile of the learner’s capabilities and adjusts accordingly (UCL Institute of Education, 2021).
Understanding Cognition vs. Metacognition in Adaptive Learning
The distinction between cognition and metacognition is fundamental to understanding how adaptive learning platforms can effectively support learner development. Whilst cognition encompasses the actual mental processes involved in acquiring knowledge and understanding—such as reading, memorising, problem-solving and applying concepts—these are merely the direct learning activities themselves. They represent the ‘what’ of learning, the basic mechanisms through which we engage with new information.
Metacognition, however, operates at a higher level of awareness, often described as ‘thinking about thinking’. It represents our ability to understand and regulate our own thought processes. This includes planning how to approach learning tasks, monitoring our comprehension and progress, evaluating the effectiveness of our learning strategies, and engaging in self-reflection about what works and what doesn’t. When a learner pauses to consider whether they truly understand a concept, decides to switch revision methods because their current approach isn’t yielding results, or recognises gaps in their knowledge, they are engaging in metacognitive processes.
This distinction becomes particularly relevant in the context of adaptive learning platforms. Whilst traditional e-learning systems might only track cognitive outcomes—such as correct answers or completion rates—more sophisticated platforms can capture metacognitive data that reveals how learners approach their learning journey. This deeper insight enables these systems to understand not just what learners know, but how they think about and engage with the learning process itself, allowing for truly personalised adaptive support.
[Chris Suggestion] Data transformation as a paradigm.
A natural follow-on to “digital transformation” for enterprise organisations has been “data transformation”; essentially the big brother of the transformation we are advocating for. After a decade or more of fighting for budget and stakeholder buy-in in their organisations, data leaders mostly found success with the following three-stage process (which can often be seen in definitions of data maturity):
- Starting out.
- Developing data-first thinking.
- Leading with data
The first stage generally involves getting leadership buy-in (most successful data transformations are led from the CEO’s office) and proving early value. This usually involves buying in existing solutions to allow the team to move fast, moving data into one place and solving early (hopefully easy) problems as a way to show that they are on the correct path. Moving fast always comes at a cost, however, and in this case it is a combination of high-interest technical and organisational debt that teams will need to pay back later.
Stage 2 is more of a hearts-and-minds operation – teams will have found their early customers in their organisation, and will now need to use these early success stories to win over less enthusiastic users. At the same time, they will need to start paying off the debts that they took on in phase 1, and establish more formal practices to help them adapt their approach to the particular structure of the organisation. Cookie cutter won’t cut it from here on (forgive the pun!).
The final phase is really about using your data maturity to do things that other organisations simply can’t do. The team at this point have encouraged a generation of organisational leaders to think in a metrics and outcomes driven way, and have adapted their infrastructure to the specific nuances of that organisation. In a very real sense, the data pipelines at this stage will encode the organisation and how it is run. The time has come where organisational change can be effected by virtue of the way in which the organisation measures success.
With this framework for data transformation in mind, perhaps there is a way we can adapt it to the specific case of L&D and HR transformations?
The Silo Problem
In the data world, fragmented systems are the cause of so much organisational heartache, and the low availability of data also hampers the capacity for orgs to use their best internal material for adaptive learning. This can be as simple has having the problem that everyone has their own spreadsheet of data without organisational coherence, or as complicated as having multiple silos of data in different platforms that are not connected.
1. Multiple Platforms, Fragmented Data
Many organisations run separate systems for HR, performance management, and L&D, each storing data in incompatible formats or behind firewall restrictions (CIPD, 2022). The AI that powers adaptivity may only see data from the LMS, missing crucial indicators such as real-time performance metrics or job-role changes from HRIS.
2. SCORM Data Limitations
The SCORM data specification, which many legacy learning systems rely on, only captures basic completion data and timestamps. This limited dataset provides insufficient insight into how learners actually engage with content. Without richer behavioural metrics, L&D teams cannot effectively analyse learning patterns, identify struggling learners, or adapt teaching strategies in meaningful ways. The binary nature of SCORM completion data (finished/unfinished) fails to capture the nuanced interactions that could inform truly adaptive learning experiences.
3. Legacy HRIS Inflexibility
Older HRIS systems frequently lack APIs or robust export functions, making data integration complex. As a result, L&D often cannot feed performance data (e.g., sales figures, customer feedback) into adaptive learning algorithms to create a complete learner profile.
4. Organisational Mindset
Some departments may not prioritise data sharing or worry about privacy and compliance. Without cross-functional collaboration, attempts to unify data fail to gain momentum.
Understanding True Personalisation vs Basic Customisation Whilst many L&D platforms offer “personalisation” through basic demographic or role-based customisation (e.g., showing different content to managers vs individual contributors, or adapting language by region), this is fundamentally different from the deep personalisation achieved by social media platforms. Social platforms analyse vast datasets that include behavioural patterns, interaction history, time-based engagement metrics, content preferences, social connections, and platform usage patterns. Corporate learning departments hoping to achieve similar levels of personalisation must recognise that AI-driven adaptivity depends entirely on the breadth and depth of collected learner data. Simple demographic segmentation, whilst useful, cannot match the sophisticated personalisation possible with rich behavioural and metacognitive datasets. The gap between basic customisation and true personalisation remains a critical challenge for L&D teams working with limited data, such as SCORM-level data.
Solutions and Strategies [If we adopt the data transformation as a paradigm, we can flesh this out]
- Establish a Unified Data Architecture: Collaborate with IT to create a central data lake or adopt a middleware solution that harmonises data from multiple systems. This is the technical backbone of adaptive learning.
- Establish Data Literacy and Operational Discipline: Ensure that every employee who works with data understands the need for interconnection, interoperability, taxonomies, hierarchies, and data management throughout data workflows, and adheres to a common set of practices that work for your organisation.
- Secure Leadership Buy-In: Demonstrate the ROI of adaptive learning pilots to executive stakeholders, showing how deeper data leads to more accurate training and skill development.
- Privacy and Compliance: Work closely with legal and compliance teams to ensure appropriate safeguards are in place, especially if capturing sensitive or behavioural data. Transparent communication with employees on how their data is used is critical to maintain trust (ICO, 2022).
The Importance of Metacognitive Data
When employees solve problems, reflect on their learning process, or share feedback about their confidence levels, they generate metacognitive data. AI can leverage this to identify strategies, misconceptions, or recurring patterns. Over time, the adaptive system does more than just push or pull content—it fosters better self-awareness and learning habits (Brown, Roediger, & McDaniel, 2014).
Conclusion
Adaptive learning represents a major leap forward for workplace L&D, but only if organisations can tap into a rich well of learner behaviour, performance metrics, and metacognitive data. Siloed platforms and inflexible HRIS solutions remain a key barrier, preventing adaptive AI systems from reaching their full potential. Good data literacy, practices, and operational discipline as part of a data strategy are key to success, not only with adaptive learning, but to succeed in the AI revolution. By investing in robust data integration, securing executive support, and clarifying data ethics, L&D can overcome these obstacles and unlock genuinely personalised, metacognitively informed training. This means not just better test scores, but a deeper, longer-lasting mastery of skills in the flow of work.
References
- Bransford, J. (2020). Adaptive Learning and Cognitive Insights. Learning Sciences Press.
- Brown, P. C., Roediger, H. L., & McDaniel, M. A. (2014). Make It Stick: The Science of Successful Learning. Belknap Press.
- CIPD (2022). People Analytics and the Future of Learning.
- ICO (2022). Guide to Data Protection. Information Commissioner’s Office.
- UCL Institute of Education (2021). Personalised and Adaptive Learning: Policy and Practice.