Predictive Learning

In todays L&D landscape, the way businesses determine who should participate in what training isnt far away from some sort of conjuring act. More often than not, the result of this is a mixed bag, and many of the L&D professionals I speak to tell me that the L1 scores (based on the Kirkpatrick model) are more often than not tending towards the lower end of the spectrum.

There are typically two ways a business determines training participation. One is based on mandated training (usually related to promotion/growth), while the other is nomination by the business manager. Both of these are based on picking up from a ‘menu’ of available programs, and neither really takes into consideration the actual learning needs of the individual.

This is where the idea of predictive learning comes in. The idea here is simple … today, with the technology available to us, especially in the Big Data/Analytics domains, the data about what has worked in the past in what context is available to the organization in a large scale. This data is available based on training, HR, and operations/business data. This rich data can be leveraged to determine what is the best training solution which would likely work in a particular employee context. Like Big Data, this neednt look at the reason (or connection) between cause and effect, rather, look at the linkages as they have been seen in the past.

An important aspect of this picture is that this shifts the focus from training and learning, and from L&D to the individual learner, and makes the entire process people-centric.

One concern with this, though, could be that the outcome of the requirements could be way too granular, and too tailored to individual needs, so as to be unviable from the delivery perspective. More about this later …


Conversation Context

If you are on WhatsApp, you are probably on a host of groups, which you have, in all probability muted. Ever seen a pattern in the conversations in these groups? If someone started a group, and you connect with old friends after say 20 years, the first few days are about exchanging notes, and reminiscing about old days. After that the lull comes in the conversation, and this is when most groups become about forwarded messages or jokes.

Friends from college have a group, which is the place for college-style conversations. After a while, the conversations turn towards politics, and discussions about the world, pretty much the way it used to be back in college (boys don’t grow up, remember?!). However, due to a number of reasons, we decided to start a separate group about political/intellectual debates. Yes, intellectual indeed, even if I myself say so. Over time, the original group, which all of us are still members of has become a group for forwarded messages.

The reason is simple … context. No conversation can happen between two people without some modicum of a shared context. Take the context away, and the conversation can’t last. As college friends, we have gone different ways in our lives. However, there is a strong shared context of our time together at college, but beyond that, the shared context is that of the world around us. And hence, these are the two topics on which conversations can sustain.

In other words, context is key.


School Exams

There is, for obvious interest, quite a bit of interest in the subject of subjects, or, rather, in the subject of examinations on those subjects. The newspapers recently reported that Central Advisory Board of Education has recommended re-introduction of the class X exams. Another subject the article talks about is the policy of student detention based on exam results.

This brings us to a basic question … what is the purpose of exams? While there is definitely a need within the education system to assess achievement of learning objectives, the problem begins when exams are seen as a mechanism to weed out students who may not meet the criterion of meeting learning objectives. If the intent is to ensure that students learn the things they are supposed to, would studying the same thing again help a student understand better than the first time? This is akin to repeating something in the hope that just by repeating it, the other person will understand it. If the student didnt understand it the first time, isnt it more than likely that he wont understand the next time either?

Instead of having the student go through the entire year, it would be more helpful for the student if the focus was to be on topics the student was facing difficulty in understanding. A quick look at the answer sheet for the exam will give the answer. This, though, wont scale without the use of technology to support this, and today, we have the technology to move assessment in this direction.

Another aspect is to find out what we are testing. Are we testing memory of the subject, or are we testing understanding? If we are trying to assess achievement of learning objectives, we need to focus on understanding. This means the pattern of testing needs to change towards application of concepts from simple recitation of concepts, and we, as a nation, probably need an examination/assessment policy to complement the education and learning frameworks in the country.


Virtual Education

Over a period of time, I have written about how virtual education can be used to address some of the problems facing education today, namely, shortage of good faculty, and the high cost of getting high-quality education to a large number of students. Few days back (which is when I had made a note to write this, but didnt quite get round to it, maybe because of the New Year holidays, but then, that would be just an excuse) I read about virtual courses offered by IITs being attended by students from other colleges (please help with a link if you could find … have been searching, but not to much avail). In this, the college from which the participating students come from decides whether they want to give credit to students for the courses they attend virtually, but if they do, then this is a step in the direction I have written about. For the participating colleges, there is, of course, the additional aspect of building the infrastructure to support the learning around these courses, in the form of labs, Q&A sessions, tests, and so on, but there is a benefit too. This would arise in the form of the content being made available to students, around which the learning then takes place.

Taking this one step further, is the possibility of building Centres of Excellence, hosted by one or two colleges, and have courses around these topics being delivered virtually by Professors from these colleges. For instance, these centres could be like Material Science, Thermodynamics, Engines and Turbo-Machines, Operations Research, and so on. Based on mutual agreement, specific colleges could be identified as CoEs for each of these, and courses delivered on these areas by Professors from these colleges, with students participating from all participating colleges. This would give the benefit of standard, high-quality education being delivered to students across the country, without necessarily replicating faculty skills at multiple locations, and enhance access as well as address cost issues.

From here, the individual colleges could take over, in supporting the learning process around these courses. This would be in the form of Q&A, homework assignments, labs, case studies, projects, exams, and facilitating cross faculty-student collaboration. In this way, the participating colleges get access to world-class content, delivering colleges build a source of revenue, and there is a symbiotic relationship between the two. The participating colleges give credits for these courses, manage the learning process, and give degrees based on their own criteria.


TM & SCM – Contd

Continuing from this post, I was thinking about more details about this parallel between Talent Management and Supply Chain Management. The first principle, from which I am trying to derive things here is that in both cases, there is a demand (in one case for talent, and in the other case for products), which needs to be met, and frameworks or processes put into place to match supply with demand. With products, the source of demand is simple to visualize. Not so with talent. So lets begin by taking a look at that.

The need of strategies, processes, and practices in the organization is to meet the business vision of the organization. To meet this vision, some work needs to be done by some people, and therefore, there is a need for people, equipped with the talent to do this work. So, the demand for talent arises from the work to be done to meet the strategic goals of the organization. Add to this the fact that there is specific talent available within the organization, and from there, its a question of trying to match available talent to the demand for talent, and based on this, determine what talent is required (in which area) to meet this demand. The supply of demand comes from employees, contractors, applicants, and L&D. I say L&D because learning is one way for creating talent supply to meet the talent needs of the organization.

Having said this, the basic concept which is the core for SCM is the concept of the part number. This is the unique identifier which tells anyone across the supply chain which specific material or product is being talked about. There needs to be a concept similar to this, something which uniquely identifies the attributes of the talent required (somewhat like part number which uniquely identifies the specifications of the material being spoken about). Different organizations meet this requirement in different ways. As you will read here, IBM solved this with the concept of JRSS, the Job Role Skill-Set, which is a composite of the job role, the role that an individual performs, and the skill sets that the individual has. This is the common identifier which can uniquely define what talent is being spoken of in the talent planning process.


TM & SCM

A discussion I was having the other day got me to think about how Talent Management is based on principles which are analogous to other functions. And this brought me to the idea of the similarity between Talent Management and Supply Chain Management, in terms of principles. If we look at the essence of Talent Management to be about bringing the right people to the right roles at the right time, then we can, from there, start looking at the essence being to match the demand for talent with available supply, and building supply pipelines where there is a shortfall.

To begin with, one of the major conconers for organizations is uncertainty. If things were fairly certain, then there wouldnt be much to be gained by trying to manage talent, because things would be running pretty much the way they are running. The sources of uncertainty are many, but thats for another time. This uncertainty results in the need to identify, based on the organization strategy, and operating plans for the coming years, what the organization’s talent requirement is going to be. Given uncertainty, there is also the need to identify how good this estimate is. This is analogous to demand planning, where the need is to estimate how much demand the organization would need for which products, and the amount of uncertainty (sometimes measured in terms of probability) associated with that demand. Based on this estimate, one can arrive at the talent required to meet the strategic and operational plans.

With this forecast as the baseline, one can then look at the talent existing in the system. This may be the talent pool which is ready for the roles for which they are required. This is akin to the gross-to-net calculation which is common in all material planning (MRP) systems. At the same time, one also needs to try to identify how many people the organization wants to, or can, develop from within, to meet these talent requirements, and from here, derive what the hiring plan looks like. This is quite akin to make-or-buy decisions material planners regularly have to make, keeping in mind available resources to make. At the same time, this serves as the input to defining Development Plans, which is akin to creating Work Orders to meet the build requirements.

Traditional succession planning is about identifying which individual should be doing what role some period down the line, but this is problematic, given that after that period of time, either the person may not be with the organization, or the role envisaged may not be part of the role directory of the organization. So, instead of looking at an individual job and a particular person, one can look at a job family to be fulfilled by a talent pool. This is analogous to product-family or product-category level planning, because forecasting, and therefore planning at the aggregate level is more accurate.


Blended Education

blog post I had written recently was what I was reminded of when I read this one about the Blossoms program, reason being that this was quite the type of delivery mechanism that I was talking about.

The idea here is simple … video recordings of lectures by a panel of expert teachers which form the backbone of education delivery across schools. This enables standard education delivery, while at the time making sure the best teachers are available to deliver classes to students across cities and villages, including in places where these top teachers would typically not want to go. Follow up this lecture with interactions at the classroom level, where the teacher running the class builds up on the video, and takes the students into interactions to discuss the content delivered, and ensure understanding of this content to all the students.

This logic of interactions can be extended to lab exercises too, as well as project reports, where the theoretical concepts are delivered in an electronic way, while the application of these concepts, including lab experiments, and the discussions among students are conducted face-to-face.

Needless to say, this method can also be extended to the L&D domain. The idea here being that with this mechanism, the L&D team can bring experts from across the world to the desktop of the learner, but the learning interaction doesnt necessarily end there (which it would in the classical e-learning method, which is still a dominant method in most organizations), and is extended with the learning interaction being extended to face-to-face (now face-to-face here could either be an actual face-to-face session, or a virtual face-to-face session), where learner interactions can happen, and learners can be presented with illustrations, case studies, and more detailed inputs, including, very importantly, inputs which are company-specific. These, of course, could be applicable both in traditional one-to-many training interventions, or in one-to-one coaching methods.