Why Data-Driven Success Demands More Than Quick Wins With Vikas Mahajan

While the quick wins are valuable, they should be part of a bigger ecosystem that scales with the use cases.

The pharmaceutical industry is undergoing a significant transformation, fueled by the integration of AI and data-driven technologies. These innovations hold the promise of revolutionizing everything from personalized medicine to drug discovery and patient engagement. However, amidst the excitement surrounding these advancements, leaders must be cautious of the risks posed by jumping into AI adoption without fully understanding its potential or limitations. In this interview, Vikas Mahajan, Head of Data and Analytics at Indegene, offers valuable insights on how pharmaceutical organizations can navigate the complexities of AI, avoid common pitfalls, and ensure that their AI investments align with long-term business goals.

With over 20 years of experience in data strategy, AI, and machine learning, Vikas has led numerous client-facing teams across diverse industries, with a particular focus on healthcare and life sciences. He has worked closely with organizations to develop AI frameworks that foster innovation, drive business value, and address the sector’s most pressing challenges.

Key Highlights:

Avoiding AI Hype: Vikas stresses the need for pharma leaders to focus on the real business value of AI and avoid getting caught up in the allure of trendy technologies that may not yield meaningful results.

Data Quality is Key: According to Vikas, the foundation of successful AI initiatives lies in high-quality, well-governed data. Without accurate and comprehensive data, AI-driven insights cannot be trusted to drive informed decision-making.

Tackling Confirmation Bias: Vikas highlights the critical importance of addressing biases within AI models, ensuring that they are trained on diverse and representative data to avoid skewed results, especially when it comes to patient care.

Driving Hyper-Personalization: Vikas discusses how AI can enable hyper-personalization in patient care, helping pharmaceutical companies use integrated data to create more targeted and effective treatments tailored to individual needs.


Kashyap: Hello and welcome, everyone, to the next episode of the AIM Media House podcast, Simulated Reality. Today, we have the Head of Data and Analytics at Indegene, Vikas Mahajan. Vikas, how are you doing today?

Vikas: I am good Kashyap. Thank you for having me for this discussion. This conversation is truly close to my heart, as it addresses real challenges and aims to deliver hyper-personalized customer experiences. I’m looking forward to connecting with you now.

Kashyap: Before we dive into the latest developments, you mentioned that you’ve spent almost two decades in Pharma and life sciences. Was that a path you chose out of passion, or did it happen by chance? How did your journey in this industry unfold?

Vikas: From a passion perspective, truly everywhere I go, I see through the patterns, the data, how the insights are coming, and I ask myself, ‘What’s next?’ If I’m traveling, I’ll say that I notice these ads that are showing up on boards and wonder how we can do it differently—how to make it more personalized for each customer. That’s the passion I have.

So, whether I’m walking over a coffee or tea, or at a certain outlet, I think about how you can recognize that this person, this customer, needs to be acquired by a different outlet altogether, through real-time messaging, offers, and discounts to make it more personalized. That’s always been on my mind—how things can be done differently, analyzing the data, recognizing the patterns, and making it happen in the real world.

Kashyap: You’re passionate about identifying patterns through statistical analysis of data. In the Pharma industry, we’ve seen rapid adoption of technology in recent years, especially in areas like drug discovery, with companies like AlphaFold making headlines. What are some of the latest and most exciting developments in the commercial side of Pharma? From your work with leaders, what are the state-of-the-art technologies and use cases being built that you’re most excited about?

Vikas: Pharma as an industry is one that’s always solving real-world challenges and making an impact globally. There are multiple areas pharma is focusing on, and if you talk about Gen AI now, what we see through studies is that we are expected to produce somewhere between 60 billion and 110 billion of annual value across the pharma value chain. Indegen’s own report showed that 62% of Pharma leaders have increased their investment in AI and ML-based applications over the past years. AI is being used in a wide range of pharma processes, including clinical trials, personalized medicine, sales, marketing, and innovative therapy solutions.

However, it’s also true that people have only scratched the surface. Our own research shows that one in three leaders are skeptical about the value they expect to generate from AI and machine learning applications. So, my mantra for determining whether AI is fulfilling its promise is straightforward. It hinges on how we define the right metrics for success. Let me give you an analogy: A strike rate of 40 in test cricket is often considered good for a batsman, and we all know that. But what do we consider a 40% strike rate in drug discovery, HCP segmentation, or campaign effectiveness? Would you consider it a good strike rate in a T20 for the batsman? The point I’m trying to make here is that we should redefine the metrics for AI success to reflect incremental, transformational gains, rather than aiming for a perfect score every time.

So, as I mentioned, this applies to clinical trials, drug discovery, segmentation, and campaign effectiveness. With pharma being so stringent on regulations, privacy protection, and global markets, this is where AI plays a pivotal role. But it has to be evaluated in combination with factors like efficiencies, reduced costs, scalability, improved patient outcomes, and the generation of novel insights. So overall, that’s where the fun element happens—ensuring we are using the technology to enable our customers and make a real impact. 

To do all this, you need a strong foundation: high-quality comprehensive data. You need to ensure that data quality is in place and that the data available for training and validating AI models has been validated itself. Without that robustness, even the most sophisticated algorithms will produce suboptimal results. That’s why I always tell my clients to prioritize strong foundations for their AI initiatives. We are investing in data quality, governance, and infrastructure. These become part of being a trusted partner, aligned with your business goals and objectives. That’s why it’s very important to be wary of shiny objects without strong foundations.

In case of poor-quality, biased, or incomplete data will often lead to incorrect predictions, misleading insights, and failed implementations. Leaders need to carefully vet AI solutions, focusing not only on their immediate capabilities but also on their data sources, quality controls, and how they can interpret the insights that come through. Let me give you an example: In one of our engagements with a leading Japanese pharma firm, we worked on personalizing engagements for HCP and boosting prescriptions for a kidney disorder drug. The client needed to proactively drive better engagement with the right content at the right time. However, accessing data and insights to make this possible posed a real challenge. Additionally, the lack of a solid and explainable recommendation engine set them back in terms of providing reps with reliable, high-contextual recommendations for their HCP engagements.

That’s where we saw adoption of AI and DNS going down. So, we extensively engaged with the sales, marketing, field teams, and their analytics team to evaluate the brand’s readiness and understand their strategies for increasing adoption rates, including change management. We incorporated an explainable AI model to provide transparent insights into the decision-making process, thereby instilling a higher level of confidence in the recommendations provided by the engine. These are the kinds of things, Kashyap, where we are seeing real impact for the client—keeping things jazzy, but ensuring there is a strong foundation behind it.

Kashyap:You mentioned avoiding flashy tech, which is a key concern for many organizations. Often, executive leadership hears about the latest technologies like Gen AI and wants to jump on the bandwagon, but these technologies don’t always bring real value. From your experience as a data and analytics consultant, how do you advise companies to ensure that the technologies they adopt are realistic and aligned with business needs? How do you guide them in educating executive leadership and fostering a dialogue about the practical applications of these technologies?

Vikas: We all have to agree that with the innovation in AI, the life sciences industry is at the cusp of a major transformation with emerging technologies we are poised to address some of the biggest challenges we currently face, especially in healthcare data analysis, to unlock clinical, operational, and business values. Companies know the potential is there—understanding patient history, tracking product performance, preparing for sequential and parallel launches, and even improving customer personalization rather I would say, moving towards hyper-personalization. But in practice, it’s not that simple. Finding the real-world data sources, making sense of unstructured data, and integrating everything into a clear, unified view of the customer journey can be incredibly tough across the life sciences value chain. These hurdles often stand in the way of adopting the winning strategies at the right time, especially when scalability and long-term brand performance are on the line.

But what I’m really excited about, and how I go about consulting our customers, is how AI is enabling end-to-end solutions that completely transform the way we approach these problems. With the rise of hyper-automated ecosystems powered by AI, LLMs, RPA, NLP, and NLQ, we can now achieve a seamless system that enables integration, management, quality controls, and transformation of large amounts of raw, unstructured data into one unified view. This end-to-end integration is key because it accelerates the journey from data to insights, allowing us to achieve both speed and precision.

Now, imagine having access to data across various streams—sales, CRM systems, claims, EMR, reimbursement channels, and even digital engagement touchpoints. When all this data is integrated seamlessly, the insights we gain from it are powerful. It’s transformative. You can now map the entire patient journey from initial diagnosis to treatment and follow-ups, while also tracking disease trends, prevalence, and outcomes across different populations. You can analyze HCP prescription behaviors, segment patients based on demographics, morbidities, and assess treatment effectiveness by correlating patient outcomes with prescription data as well.

Even the impact of digital engagement on both patients and HCPs becomes clear. With this level of integration, you can make more informed strategies. This is what truly excites me about the future of life sciences—the power of an AI-driven ecosystem that brings together all these touchpoints across the value chain. It’s going to reshape how we handle patient data, make it more hyper-personalized, and improve outcomes on both the clinical and commercial fronts

Kashyap: Let’s dive deeper into a common challenge in industries, particularly in pharma—the issue of confirmation bias. It’s often cited as a major obstacle in decision-making, especially in traditional sectors like pharma where established processes and beliefs are hard to change. Data leaders are aware of this and try to address it, but many business leaders still rely on confirmation bias, seeking data that simply aligns with their preconceived notions. As a data analytics leader consulting these organizations, how do you help them overcome this bias? How do you guide leaders to make data-driven decisions and challenge those pre-existing ideas?

Vikas: Okay, so I think, firstly, you raise a very valid concern, Kashyap. Confirmation bias is a significant issue. We all suffer from our own biases, and for AI systems, while they are powerful, they can unintentionally reinforce this bias if they are not trained on the data sets correctly and, rather, trained on biased data sets. After all, whatever is happening, AI systems are human creations, at least for now, and that’s why they can often mirror the creator or the society. I came across a research review which found that clinical AI models are mostly trained using data from the US and China, and these countries have advanced AI infrastructure as well. This heavy reliance on their data creates a risk of bias and also limits the model’s ability to work accurately in regions with different demographics. So, the key here is not to ignore the risk but to confront it directly. Pharma leaders need to accept that AI models, like human decision-making processes, are prone to bias, and the first step toward mitigating it is to acknowledge its presence.

So, I think addressing the bias in AI is not only the right thing but a smart thing for a business, and the stakes for business leaders are high. If we don’t address this confirmation bias, the results can be catastrophic, especially in a field like pharma, where decisions can directly impact your patients’ safety and outcomes. So now, by recognizing the confirmation bias, leaders can proactively build systems to minimize its impact through better governance, models transparency, and rigorous validation processes. That’s why I believe that healthy skepticism, if you will, is good for the overall progress and advancement of AI/ML technologies.

In fact, if you see through the healthy skepticism towards AI, it’s not only justified but necessary at certain times. Just as early skepticism in the aviation industry or even in the early days of the internet led to most engine safety protocols and design improvements, questioning AI’s validity can lead to better systems in the future.

This skepticism drives continuous evaluation of AI models, pushing developers and regulators to scrutinize data quality, algorithm fairness, and the transparency of it. It can ensure that AI becomes a more reliable tool for decision-making in pharma, minimizing the risk of bias or inaccuracy altogether. Now, to your question, what can pharma leaders do to address this confirmation bias? To start with, leaders must adopt a comprehensive strategy to ensure that AI solutions are effective and unbiased. To tackle this, there are several factors that leaders need to pay attention to.

First, the quality of the data. The source of the data is paramount—it’s only as good as the data it’s trained on. So using diverse, high-quality data is the key to avoiding bias or skewed results altogether. Secondly, having effective data management and governance in place is really important to ensure that the data is accurate, timely, and robust. Strong governance frameworks help improve the overall efficiency and scalability of AI initiatives.

Third, keeping humans in the decision-making loop can help identify potential biases in your AI outputs. This could mean having clinicians review AI-generated treatment plans to ensure they align with real-world patient needs as well.

The fourth point is that leaders should align AI goals with real business objectives and patient outcomes so that genuine needs are being addressed, rather than delivering insights that align with preconceived notions. Lastly, the model should undergo continuous testing to identify and mitigate bias before they’re deployed at scale.

So in summary, what I’m trying to say is AI-driven insights have the potential to revolutionize healthcare by enabling personalized, efficient, and effective treatments. However, it’s crucial to address the challenge of confirmation bias and ensure that we deploy them responsibly.

Kashyap: While all four points you mentioned are important, two stand out to me: the relevance of the metrics and aligning them with business objectives. There’s an overwhelming amount of data, and while it’s possible to create numerous dashboards or models, the real challenge lies in ensuring consumption and impact on decision-making. As a leader, how do you ensure that your models and dashboards are not just created but are impactful and aligned with business objectives? Is there a specific framework or approach you follow when working with your team and clients to guarantee relevance and value?

Vikas: That’s a great question, and it’s a challenge we often see as well. Leaders need to differentiate between value-driven AI, which directly contributes to business growth or patient outcomes, and vanity AI, if you’ll say. These are flashy, expensive projects that may showcase cutting-edge technology, but lack real-world evidence.

These metrics, like social media engagement or open rates, aren’t inherently bad. They serve as a baseline to understand things. It’s like cooking with raw ingredients—they might not make the meal, but they’re crucial components. Vanity metrics can be useful if they’re understood as just one piece of the puzzle. For example, if you’re running a LinkedIn campaign and noticing low lead generation, vanity metrics like impressions or shares can help you figure out where the issue lies. Low impressions might mean you’re not reaching enough of your target audience, so you might need to adjust your targeting or broaden your audiences. But if impressions are high and engagements are high, like clicks or form fills, that means your message is resonating with the audience, and you can adjust the messaging accordingly.

But the key is how leaders combine these metrics to make them more meaningful. This data should inform product strategies and drive outcomes, like customer retention, revenue growth, and personalization. Achieving that is simple if you go beyond the surface. It’s not enough to just look at high engagement numbers. You need to ask questions: Does this engagement lead to conversions? Are these interactions translating into long-term relationships? What success looks like is critical.

Secondly, context is everything when it comes to data. What might seem like a success in one context could be meaningless in another. For example, a surge in app downloads might look impressive, but without a corresponding increase in active users or purchases, it’s just a vanity metric. Leaders need to align metrics with specific business goals. Contextualizing data ensures the focus is on what truly drives performance, not just what looks good on paper. I might be fast in the field, but if I’m not running toward the finish line, it doesn’t matter.

In my experience, leaders fall for the trap of vanity metrics because they provide a quick sense of achievement. They want to look good, and these metrics are easy to understand and offer instant gratification. But the problem arises when leaders stop there—they fail to ask the harder questions and track metrics that reflect real business impact. I was recently involved in an org level initiative around continuous improvement, and while the focus wasn’t just around AI, the learnings from that initiative are equally applicable here. Leaders need to encourage their teams to adopt a culture of continuous improvement. Success should be measured by outcomes, not optics.

They should prioritize creating an environment that values deeper insights over quick wins. When I work with clients, I always tell them that the metrics that were once relevant or helpful can lose their value as market dynamics shift. What drove success five years ago may not have the same impact today. Metrics need to evolve as the business evolves. I like to call this the evolution toward actionable metrics, which make sense for the business, provide personalized recommendations, and are driven by business outcomes like revenue, customer satisfaction, and retention. This gives a true picture of success.

Kashyap: The points you raised about quick wins and short-term success are crucial. When leaders invest in AI solutions, they understand it’s a heavy investment—resources are expensive, and scaling infrastructure adds costs. But there’s pressure for quick wins. Practitioner leaders, with years of experience in data science, often face unrealistic expectations for immediate ROI, especially with emerging technologies like Generative AI. How would you advise both senior leaders and practitioners to set realistic expectations and work together for long-term success?

Vikas: So, I think very relevant questions. While short-term wins are essential for building momentum and validating the potential of AI in the organization, however, these wins, if they’re not aligned with the long-term vision, can lead to fragmented efforts and missed opportunities. So, the key here is balance—connecting these quick wins to a larger AI roadmap to ensure that these immediate successes don’t distract from achieving more transformative and sustainable outcomes. Just like learning to ride a cycle, companies need to understand that rejoicing around the block using those training wheels will not set them up if they want to go faster and cover that distance eventually.

For instance, when it comes to omnichannel reach, you can use the data sets from past behaviors to achieve short-term success. But to fully capitalize on this, you need a long-term strategy that involves using data across various campaigns, assessing effectiveness, and determining the next best actions to deliver hyper-personalized customer experiences.

While the quick wins are valuable, they should be part of a bigger ecosystem that scales with the use cases. Take our work with hyper-automated AI, for example. We integrated data sources from commercial operations, built automated data pipelines, and developed aggregated reporting based on the end-user personas altogether. And this wasn’t just one-off; it was designed using MLOps, RPA, and other cutting-edge technologies to make it scalable and future-ready for our clients, enabling use cases to be continuously added.

In fact, we’ve seen this hyper-automated AI ecosystem deliver impactful outcomes for our customers. They were able to reduce their data preparation time by 30%, and we saw a 1.5x improvement in targeting and delivering customized communications to customers altogether.

So, short-term wins—like sprints in a marathon—can give you a burst of energy and help you cover ground quickly. But the ultimate goal is to complete the entire race of it. For AI to reach its full potential, pharma leaders need to shift their focus from short-term gains toward the development of a comprehensive AI strategy that aligns AI efforts with long-term business objectives. This includes building a culture of innovation, continuously assessing how AI can deliver beyond an isolated success or an isolated DU driven focus.

Kashyap: While the tug-of-war between short-term wins and long-term goals will always exist, especially with different stakeholders, agendas, and internal politics, what are some of the exciting use cases and technologies emerging in Pharma that you are personally working on or are most excited about? For example, I recently heard a leader mention that 15 months ago, they were running 30 POCs to push forward, but now they are focusing on just four scalable use cases. How does this shift resonate with what you’re seeing in the industry?

Vikas: As I started saying that, what I have seen is that now in Pharma, from a technology event, if you’ll say so, we are seeing that there has to be an impact on the real world. There has to be an impact on your customers. There has to be an impact which matters for our customers as well. So with this happening, we need to see how your patient outcomes are driven, how your business growth is being valued through what is the value delivered insights to our customers, which are happening as well.

And as I mentioned in the last point, an integrated ecosystem approach, which is happening where now clients—the organizations—are also seeing the value in bringing everything together, rather than giving the business a siloed view, as well.

So primarily what I have seen in the past is that the data strategies were not clear. It was a major issue. The inaccurate results being treated as accurate was another major issue altogether. Then, a temptation to overinvest in tech without a clear use case was a major issue. IT infrastructure was not prepared for demands of AI and scale to the future altogether. Then finally, fragmented AI strategies across different business functions, which you mentioned, was another red flag altogether, as well.

So now what I’ve seen, which you mention as well, that bases all these major red flags, what we have seen is that there is a comprehensive data strategy into the picture, which is coming up, and I am heavily invested into ensuring that when I’m consulting my clients, I’m going to them with the proposition.

The first thing which I’m laying down is that data is the fuel that powers AI. And if the data is not relevant, comprehensive—yeah, AI won’t deliver meaningful results. Secondly, I’m ensuring that the models which we are training are built on trustworthy data. They should not lack traceability. AI needs to be giving outcomes which are aligned to the business outcomes.

There have to be established feel-safe mechanisms altogether as well. And there has to be human oversight where the results are reliable and actionable. We are also ensuring that the organizations which have separate AI strategies for each department, if you’ll say so, do not work in silos because that stops them from being fully integrated across the enterprise.

So we are now focusing on how this should be treated as a cohesive strategy, not as an isolated initiative altogether for the organization, because then the impact—the value of those insights—would be spread across the enterprise, that would be realized by the users, by the customers. And that is the strategy.

So, while now in Gen AI there are multiple use cases which are coming up, ensuring that they’re all aligning with your vision, they’re all strategic in nature, and now that you have the data, you have the relevance, you have the impact, what you need to deliver going forward is all being well thought through.

So I think that is the main thing. I’m truly excited that organizations are taking those steps. They’re aligning toward that going forward as well. So I’m looking forward to working in that direction as well.

Kashyap: Fantastic! On that note, Vikas, thank you so much for making the time and sharing your thoughts with us.

Vikas: Thank you again for having me here, Kashyap. It was a great conversation, and I’m really excited to see how AI technology continues to drive real transformation in the industry as well. Thank you.

📣 Want to advertise in AIM Research? Book here >

Picture of Anshika Mathews
Anshika Mathews
Anshika is the Senior Content Strategist for AIM Research. She holds a keen interest in technology and related policy-making and its impact on society. She can be reached at anshika.mathews@aimresearch.co
Subscribe to our Latest Insights
By clicking the “Continue” button, you are agreeing to the AIM Media Terms of Use and Privacy Policy.
Recognitions & Lists
Discover, Apply, and Contribute on Noteworthy Awards and Surveys from AIM
AIM Leaders Council
An invitation-only forum of senior executives in the Data Science and AI industry.
Stay Current with our In-Depth Insights
The Most Powerful Generative AI Conference for Enterprise Leaders and Startup Founders

Cypher 2024
21-22 Nov 2024, Santa Clara Convention Center, CA

25 July 2025 | 583 Park Avenue, New York
The Biggest Exclusive Gathering of CDOs & AI Leaders In United States
Our Latest Reports on AI Industry
Supercharge your top goals and objectives to reach new heights of success!