Upskilling is Broken: Myth, Model, or Management Failure ft. Mahesh Raja | Hot Takes and Hard Truths

Mahesh: Every employee is expected to work long hours, nine or ten hours a day, and then they’re expected to upskill themselves. That creates more burden for employees.

Nikhil: Upskilling is the biggest buzzword trend right now, especially over the last couple of years.

Mahesh: It’s a big challenge for organizations. How do you know what skill gaps actually exist? Research shows that within 30 days, almost 90% of what people learn in such programs is forgotten. When I say skill to order, I mean focus on today. When I say skill to stock, it’s about tomorrow. You can’t put a bunch of 40-year-olds in a classroom and expect them to learn without seeing value. As much as you want to put 100,000 people on a Gen AI training, the question is, can they apply that learning in their day jobs within 30 to 60 days?

Nikhil: Every time there’s a technology shift, Cloud, Automation, or now Gen AI, someone writes the obituary for Tech Services: “It’s over. The industry won’t survive the next wave.” And yet, the best teams keep building, adapting, and staying relevant.

My name is Nikhil Kulkarni, I’m a Partner at Zinnov. Welcome to Hot Takes & Hard Truths, a subseries of our Business Resilience podcast.

Right now, there’s no shortage of hot takes:

• Automation killing services

• Gen AI replacing engineers

• The end of large outsourcing deals

In each episode, we pick one of these hot takes and break it down with leaders who are actually living it. Honest conversations, real stories from the ground, that’s what we promise.

Our guest for today is Mahesh Raja, Chief Growth Officer at Ness Digital Engineering. Ness is a digital-native, full lifecycle Product Engineering firm backed by KKR. Mahesh has nearly 30 years of leadership experience, most of them at Wipro. He joined Ness 18 months ago and leads global sales and go-to-market

Mahesh: Thank you. It’s a pleasure to be here, I’m humbled and honored.

Nikhil: Great to have you too, Mahesh. So let’s get started. The topic for today’s discussion, the hot take we are going to discuss, is on upskilling.

There is no shortage of investments in L&D across tech services firms right now. So the question I’ll start with is: Is upskilling at scale actually working? Or are we clinging to the idea more than the actual impact on the ground?

Mahesh: Before I answer your question, Nikhil, first of all, this is an amazing topic. And the title of this podcast is Hot Takes & Hard Truths.

When I was preparing for this, I was thinking, how open and honest can I be? So I will try my best to be transparent. Hopefully, the audience can resonate with my own personal experiences as well as what I’ve seen in my professional journey.

If I rephrase your question—where do companies go wrong in upskilling? I see three reasons, actually three specific reasons where companies fail:

• Chasing buzzwords and generic trends. For example, AI skills, digital transformation. Instead of being data-driven and looking at what employees actually need to serve clients or internal needs, companies jump on hype.

• Employee burden. Employees already work nine or ten hours a day, then they’re expected to upskill after that. It becomes half-hearted. Upskilling should be mainstream, not a sidekick.

• Lack of sponsorship. For any program to succeed, you need C-suite sponsorship. There has to be a purpose, and a clear benefit for employees.

Nikhil: Absolutely. Let’s pick on one of the points you made, that sometimes it’s driven by trends rather than business demand or employee skill gaps. What can we do to ensure training programs are more aligned to business demand than to market buzz?

Mahesh: The way I look at it, first, we have this concept called skill to stock, which is pretty prevalent in the market. What I mean by that is, management teams and L&D teams want to enable workforce for the future.

But what they miss is making the workforce ready for today. That’s a big gap.

Even if today we train our workforce on an upcoming technology that’s six months out, the yield on such investments isn’t commensurate with the effort. Research shows that within 30 days, 90% of what they learned is gone.

So we need a balance between skill to stock (focus on tomorrow) and skill to order (focus on today).

At Ness, our mantra is “right skills at the right time.” We don’t force programs on our 3,700 employees until there is a clear business need, whether client-specific or internal.

Example: last year we decided to upskill our Data Engineers into the latest stacks. There are many stacks: Fabric, Snowflake, Databricks. We made a call to run specific programs around lakehouse architecture. Because if you understand the lakehouse effect and the architecture, you can adapt to whichever tool is dominant, today Databricks, tomorrow maybe something else.

Nikhil: That’s great to hear. Let’s go specific. Right now, a popular method is self-paced learning, you assign certain hours to complete, expecting things to work out. But sprint cycles, delivery pressures, changing customer demands make this tough. What’s your take?

Mahesh: Two parts:

• How do you assign achievable goals?

• How do you motivate people?

On goals: incentivization is key. For example, if you’re a Java Developer moving to full stack, you must achieve certain levels. Tie programs to career trajectory.

On motivation: show the company’s investment. At Ness, we don’t differentiate. Every employee has a fixed learning budget. Depending on role and business need, they sign up. We prefer they choose, self-motivation increases yield.

Also, adult learning is different. You can put kids in a classroom, but not forty-year-old professionals. They need to see value. If I do a Databricks Level 3 certification, it helps me deliver better.

We saw in our whitepaper last year on Gen AI’s impact on developer productivity: outcomes differed by seniority. Junior, mid-level, and senior engineers responded differently. That’s why adjacent skilling (building close-by skills) often works better than broad cross-skilling.

Nikhil: Took us 15 minutes to reach Gen AI, that’s an achievement in itself! But now that we’re here, how do we look at Gen AI holistically?

Mahesh: Two perspectives.

One: real-time learning assistance. Tools like GitHub Copilot offer contextual code suggestions, explanations, examples. They help techies become more productive.

Two: improved engagement. About 70% of engineers who went through Gen AI-infused programs reported higher engagement. We didn’t expect that, but it came out.

People are used to bite-sized, 30-second impact (thanks to Instagram, TikTok). That’s why infusing Gen AI into learning can gamify and contextualize it.

Also, adaptive learning: two people on the same team may have different gaps. Gen AI allows for real-time adaptation, customized roadmaps, personalization at scale.

Nikhil: That’s a good point. And let’s say, assume we were starting from a completely blank slate. There is no LMS baggage, there is no budget constraint. How would you then look at designing an identity program ground up? What are the top two or three things you would take into consideration if you have to design a first-principles-based learning system today?

Mahesh: I’ll give three or four points that come to mind.
Number one, like you said, it has to be very contextualized to the individual or the role. A sales engineer’s needs are very different from what a technical engineer would need from the same content.

A sales engineer also needs to know, say, how AWS is coming up with a new data platform, so their ability to understand the platform and articulate it to a client or prospect is very different from how a level-two or level-three data engineer needs to understand it.

Number two, a big gap today for organizations is: how do you know what skill gaps you have? That’s where new platforms and AI help us detect gaps, like a data analyst needing Python and Power BI, and then we curate targeted micro-courses so that in a client scenario they can actually address the problem.

These are two heterogeneous stacks, but they may both be essential for a level-two or level-three data analyst. It’s not about the platform; it’s about knowing the individual’s need.

The third one, as we spoke briefly, I’m a big believer in gamification: real-time adaptation of the content. If I take a course on Python and I’m breezing through the basics, the content should adapt: give me capstone projects, tougher challenges to solve. That adaptation has to be real time and specific. A lot of content today is meant for the masses; it’s not customized for the individual. Customization plays an important role.

So, to summarize: role-based personalization, skill-gap analysis, real-time adaptation, and customization at scale. These are the four foundations I’d set before bringing in an L&D platform or making those futuristic investments to upskill the workforce.

Nikhil: That’s a good point. And let’s say, if you are advising one of your peers today from a different Tech Services company, what is one thing that you would ask them to completely stop doing right now when it comes to upskilling, and what is one thing they should absolutely start doing from tomorrow?

Mahesh: I think I’d ask them to stop doing upskilling as just an initiative, which sounds like the antithesis of this discussion. But I’ll actually say that: reimagine what roles the organization needs for today and tomorrow, and then look at upskilling as the means to achieve that, not the other way around.

The challenge we have, and I read this in a research paper, is that many times when people go through these programs, they’re not given an opportunity to apply the learning. If they can’t apply it, the investment is wasted.

So as much as you want to put a hundred thousand people on a training, the question is: do you have an opportunity to let these hundred thousand people apply that in their day jobs within a reasonable time: 30, 60 days? Otherwise, that investment is pretty much written off.

So while there’s a lot of boardroom push to “Gen AI enable our workforce,” you have to be very clear where it actually makes an impact.

Nikhil: Absolutely.

So just summarizing what we heard today, we started with the topic that upskilling is the biggest buzzword trend in the last couple of years. It’s always been there, but right now there’s a lot of skepticism around whether we’re really doing what we need to. The answer, as always, lies somewhere in between.

Based on this discussion, hopefully people learned some best practices you shared today, Mahesh, that will help them in redefining or redesigning how they look at L&D. Thanks a lot for your time, Mahesh. I hope the audience found this useful too, it was lovely speaking to you.

Mahesh: Thank you, Nikhil. If I may just make a closing comment, I really like this format of Hot Takes & Hard Truths. You didn’t constrain my thoughts to a narrow subject. You kept it open-ended. Thank you for giving me this opportunity. I really appreciate it.

Nikhil: Absolutely. Thank you so much.