Powering Enterprise Decisions with Data and Analytics represents a transformative approach that leverages the wealth of information available to organizations today. In an increasingly data-driven world, businesses are recognizing the pivotal role that data and analytics play in shaping strategies, refining operations, and ultimately driving decision-making processes. To give us more insights on it we have Pranay Agrawal, Co-founder and Chief Executive Officer at Fractal Analytics. Pranay has spearheaded Fractal to become a globally renowned analytics company, collaborating with over 50 Fortune 500 firms to enhance business outcomes through the integration of analytics into decision-making processes. With an impressive Net Promoter Score exceeding 75, Fractal stands out for its client-centric approach. Pranay holds an MBA from the Indian Institute of Management (IIM) Ahmedabad, a Bachelor’s in Accounting from Bangalore University, and is a Certified Financial Risk Manager accredited by the Global Association of Risk Professionals.
In the interview, Pranay Agrawal will cover Fractal’s founding vision during the IT-centric era and its shift to pivotal business decision-making. He’ll discuss Fractal’s decision evolution, initial client challenges in data analytics, and the widespread trend of data-driven approaches. Pranay will touch on strategies for decision impact, Fractal’s unique differentiators, and current challenges in fostering data-driven choices. He’ll also highlight Fractal’s approach to ethical challenges and conclude with insights into the company’s future in data analytics.
AIM: Can you share the founding vision behind Fractal amidst an era dominated by IT services, and how did the transition from IT-centric data ownership to pivotal business decision-making influence the company’s inception around two decades ago?
Pranay Agrawal: We were just in the midst of wrapping up a.com venture, which was going nowhere. And so, we were thinking, what next? What should we do from here?
IT and the whole Internet were a big rise at that time, but there were many companies already, And the question was, “What could we do that is different and create value differently?” And Srikant, at that point, came up with this idea and said, how about applying mathematics to businesses? If we can apply mathematics to businesses and help them make better decisions using mathematics, that would be different. And, of course, that meant algorithms; it meant data, and that’s kind of how we started thinking about it. It’s a unique idea, it has tremendous potential for creating value. We were inspired by some of the things that we saw, such as credit scoring that was being done outside of India in the developed markets. We had personally done a fair amount of work on the capital market side applying deep algorithms to solve market-related issues and opportunities, and we figured it was possible to apply this to a wide range of applications within businesses, and that’s kind of how the idea took shape.
AIM: How has decision-making evolved for Fractal over time? Convincing clients about the value of data analytics services in the early days—was it challenging due to the less prevalent landscape, or did clients readily see immediate benefits?
Pranay Agrawal: It wasn’t an easy sell, and it also varied a lot by market. We started the initial outreach or conversations in India and Asia, and then eventually, we branched out to the US. And it was an easier sell in the US. Largely because the size gives you a certain advantage where you can start to make small decisions, and when those small decisions give you an advantage over a very large base, that totals a substantial benefit for the company.
The purpose around what we do, which is powering every decision to help our clients deliver better outcomes to their customers, the employees, their shareholders, and the communities that they operate and that purpose or mission has stayed pretty much the same over time, The method of delivering it, however, has evolved a lot. When we started, it was about using your standard traditional statistical models, linear regression, logistic regressions, and various other things to help make predictive decisions.
And then, over time, that has evolved with various things. The technology around mathematics or modelling has evolved, and AI took off in a big way in 2014. One pivot that we had was going from analytics to AI. The second big thing that’s happened over time is that the ability to store data and use data has increased tremendously. And therefore, it presented the real opportunity to move from insights and decisions to operationalizing whatever you do. So you start getting to this area of Enterprise Scale, AI-driven applications. Then we realize that if that’s what we’re going to do, it’s very important to put the user at the centre of the solution. That user could be an internal stakeholder within a company; it could be their customers, and therefore, design and behavioural sciences become very important. So, today, we are doing this. We are powering every decision in the enterprise by bringing together this combined force of AI, engineering and design.
AIM: Is the widespread belief in data-driven decision-making a trend you’re seeing across the industry? If it’s evolving, what steps is Fractal taking to empower leaders with data-driven approaches?
Pranay Agrawal: Firstly, all businesses want to make good decisions; that is just a factor because everyone knows that if they make better decisions, they will deliver better outcomes. So that’s number one. The amount of data that companies use varies by the nature of the decision, it’s not just about the amount of data decision support system. It is the process as well; it varies by the type of decisions, and of course, it varies by different companies. And it is possible that you are a digital native e-commerce company. And that’s how you’re born with technology, data, and everything driven that way. So it is a very natural place for you to operate in. It could be different from another company. But what we find is that, three categories of decisions, that companies can broadly make. One is strategic. These are decisions that are high impact and very low in frequency. That could be things such as a new product launch entry into a new geography, an M&A, or a brand re-statement. So, these are very high-impact decisions. These are infrequent decisions. The cost of going wrong is very high, and there isn’t enough frequency of these decisions for us to make algorithmic-based decisions.
So, a lot of data analysis can go into it. But then there is a lot of judgment that also needs to be applied. Okay, then you think of tactical decisions, which could have a lower impact than strategic decisions. And there’s more frequency around these decisions, which could be things like, “Okay, how many units of this product? Should I manufacture? It could be something like, Okay, where should I open my next warehouse? Where should I open my next retail store? Or what price range should I operate in and stuff like that.” So, these are medium-impact decisions, and the frequency is lower. But there is the possibility of supporting human decisions with algorithms over here. So you people still have to make those decisions. However, algorithms and past experiences can start to support those decisions. Then you have operational decisions, and these are decisions that are very high in frequency. Each one of those single decisions by themselves has a low impact, but collectively, if you can improve the quality of the decision-making around thousands of thousands of those operational decisions, you can drive to a much better level of outcomes, and these would include things like, Where can I place my next digital advertisement? What price should I offer tomorrow, given something that’s changed in the market and those kinds of things? To whom should I make the next offer, and whom should I call to make a collection? And those kinds of decisions. And these are decisions where the largest amount of automation and algorithmic decision-making can occur. You see that across companies, they all treat these decisions differently.
AIM: Is the strategy about starting with low-impact, high-frequency decisions initially, gradually moving towards higher-impact and frequency decisions for companies?
Pranay Agrawal: Companies as they start to experiment with new technologies, and I mean that holds true for AI and everything around algorithmic decision-making. Yes, there is a clear need to experiment in areas with a low cost of failure. And things that can be very easily reversed. So, as an example, one of the things that we are telling our clients today in the area of Gen AI is that we understand that there are tremendous risks; I think there could be a bad customer experience there could be, brand risk, There could be a risk of inadvertently, doing something that’s not ethical, or it’s biased. One way to deal with that is to learn and get better before we go to these higher-impact decisions. Maybe we should start by focusing more on inward-oriented applications where our risks are lower. As an example of automating, a range of things may relate to documentation. I need to do underwriting, and my underwriting process is very manual and tedious. It involves collecting vast amounts of data from different places. Putting it all together and then evaluating it all. Can we use Gen AI to collect all that data, populate all the documentation, and smooth out that decision-making process? So just an example of ways in which companies can start to do this in a low-risk manner, learn and then start to make higher and higher impact decisions using data and algorithms.
AIM: What sets Fractal apart in its approach to empowering enterprises with data-driven decision-making compared to other data science service providers or product companies focusing on similar objectives?
Pranay Agrawal: I think it really stems from our first value at Fractal, which is always putting the client first. Our thinking here is that we have to make decisions backwards. Not data forwards and not data and technology forward. One approach is to say that I have all of this data and technology. Let me see What I can do with it. And that will lead you down one path. Another approach is that our mission is to help power every decision in the enterprise. It comes from the very clear thinking that if we help our clients make better decisions, they will deliver better outcomes to their customers, employees, shareholders, and the communities they operate in. So, what are these critical decisions that our clients need to make? What are the important goals, problems, and opportunities that our clients face? How do we support those? How do we help them take better action supported by better decisions, and then you move backwards from there? For these decisions, these are the kinds of insights we need, and therefore, we need such kinds of algorithms. These are the kinds of data sources that we need to support those algorithms. And now, how do we bring it all together while keeping the user at the centre of the solution? So that is the core secret. Decision backwards, user-centered and bringing AI engineering and design together.
AIM: How does Fractal equip its data scientists to effectively consult with seasoned professionals within large organizations, convincing them of better solutions despite their extensive experience, beyond solely relying on data?
Pranay Agrawal: Firstly, what we do is that we are very clear believers and implementers that big problems cannot be solved by just one skill set. Data science and AI are just two of the skill sets that are required. Ultimately, we have to create an impact. And that means that we need to understand the client’s business. And, of course, we will not know the clients’ businesses better than they do in all situations. But then we know the business well enough to ask the right questions. to be able to understand the problems and frame and reframe the problems. That is one part of the skill set that’s important. And when we say client first, it does not mean that you do exactly as the client says. To create real value for the client, you have to ask questions, and you have to enable them to reframe their asks. So I think that is a very critical part of the skill set, the ability to consult and to ask questions and understand real business issues. Once you have that and you have framed the problem and framed a solution structure around that, then you apply the different skills to different parts of that solution creation journey. Some of it might be around creating the underlying data infrastructure or getting the right data, cleansing that data, and ensuring that data is constantly feeding into the algorithms. Some of it is around data science to create those algorithms. Some of it would be around creating those front-end applications through which the users can consume the AI and the data without having to see any of the complexity and messiness behind it because, ultimately, the users don’t care. They want a very simple way in which to use and to make their decisions.
AIM: What current challenges does your team encounter in 2023 while fostering data-driven decision-making within client companies, considering the prevalent acceptance of this approach in the market?
Pranay Agrawal: So firstly, I mean again, I’ll put it at two different levels. One level would be that if I put at an organization level, which is that organizations that have created the mandate and that I’m moving forward, what might differentiate organizations that are doing this more purposefully and aggressively versus others that may be slow to adopt, and I think that comes down to, one is really, the vision and imagination. So right at the top, does the organization have the leadership, have they provided the vision around implementing these technologies to drive better outcomes in their business context and then equally, what are the fears within the organization, and how are they addressing those fears? Whether those are fears around privacy, fears around doing it ethically, or all of those kinds of things, those are two aspects that are very critical and give the organization the permission and the vision to do more or not. That’s one. When you come down to specific problems, one of the most important things is designing the problem or framing the problem right up front. That is the really important step without which you could be down the path of spinning a lot of wheels around data and technology without a clear purpose in mind. So agreeing on that is very important. And even after doing this for several years and having several successful implementations now and then, you’ll run into situations where you’ve been doing stuff for several weeks to find, he’s somehow missed out this thing right at the beginning, which, if this had been agreed upon and discussed at the beginning, there might have been a different path to take. So that is one aspect.
Algorithms are very data-hungry. So, that is an issue, and I don’t call that a challenge or an obstacle. It’s just a genuine thing that we have to address, and everyone keeps getting better at addressing it. As we continue to work on it, we are ensuring that we are providing good, clean data and continuously finding new ways to enrich these algorithms with different data sources. And the third is that this is an ongoing issue. It is going to be around talent. People who know how to develop the technology and people who also have the ability to deploy and use the technology. So those are some of the key things.
AIM: Could you elaborate on the unique approach Fractal has taken by appointing a Head of Responsible AI and shed light on their role and accomplishments in addressing ethical challenges in the realm of data science services?
Pranay Agrawal: What I laid out, number one, is that vision and addressing the real fears and concerns. And one of the fears around ethics and privacy and all of that stuff. The next step is to define the problem statement and have the right data and talent. In terms of ethical AI, we figured that this is going to be important, and there will be more and more requirements not just from governments but also from people as citizens. And we felt it was the right thing to do and get ahead of the game. So, we have created an ethical AI framework which is focused on things such as transparency, explainability, and elimination of bias. So, there are various components to this. What we do is that we internally assess how every piece of work is clearing the AI framework, the responsible AI Framework that Fractal has created, and then we are also running the sole responsible AI framework workshops for our clients. So, there are clients who get educated about this. Then, in some cases, we’re helping them implement this. And then, in some situations and partnerships with clients, we’re also certifying projects as responsible AI certified. It is still the early days, but we have created the frameworks, the technology, and the education material around it, and we’re working with our clients to make this an integral part of all AI work.
AIM: What are your closing thoughts on Fractal’s future prospects, particularly concerning data analytics and the broader landscape? What do you foresee in terms of its growth and advancements?
Pranay Agrawal: Look, we are very bullish and very positive about the overall space but not just from the standpoint of it being a great business opportunity. I think overall, if you look at the impact that AI can have on the wellbeing of humankind and how we can create a better world across a whole host of areas, we feel very bullish about the impact of AI and technology.