Building Trusted Self-Serve Analytics for Better Commercial Decisions with Ashutosh Katiyar

Hello and welcome to CBO Magazine interview series. I'm Subruto Chatterjee, Global Growth Partner for Life Sciences at Tiger Analytics. Tiger Analytics is a global AI analytics consulting firm that helps enterprises turn data into business value through advanced analytics, data engineering, AI, and decision intelligence solutions. Today, I'm delighted to be joined by Ashutosh Katya, a good friend of mine, Executive Director of Commercial Strategy Insights and Analytics at Regenron.

Ashutosh, thank you for joining me. Thank you, Subharto, for having me and having the conversation. Just a brief introduction for me. I lead the Insights Analytics function for Ophthalmology team at Regenron.

I also am serving in an entire capacity to lead a lot of commercial AI initiatives across the therapeutic areas. I've been at Regenron for almost eight years. Just so that everybody knows, Regenron is a biotechnology company that operates in Tarrytown, New York. And we serve a lot of different kinds of therapeutic areas, immunology, oncology, ophthalmology, rare diseases, and several things in our pipeline coming up.

so my question is you have pushed hard for self-serve analytics across various therapeutic areas so what was the specific problem that you were trying to solve and most importantly why was it urgent yeah no really good question um i think sort of continuing what i was talking about before right um spending way too much time on answering a lot of what's right so the team And I think, you know, as we think about the sales of analytics, right, with some of the user interface and dashboards that we want to create, right, and also, you know, with the advent of Gen AI, right, what has happened is that you can query the data, you know, very, you know, simply and in a very more agile manner, right?

And so you don't have to kind of write SQL queries every time a question comes up to kind of ping the data warehouse and get the results, right? The idea is to start setting up platforms that are serving that need of connecting the sort of like end business user to really the data, right? So you are sort of really short-circuiting that right entire stream and allowing you to really then focus a bit more on more strategic value at work, right? Thinking about mapping the decision framework, right, that a stakeholder might need, right?

So you are kind of going more from answering a lot of work questions to kind of really now being that strategic decision maker or being a strategic advisor to your business unit. And then helping them make decisions that are really going to be critical for the business, right? So in that context, cell-cell analytics really serve that foundational layer that free up your capacity and time to focus on this higher order stuff. And I think the Gen AI, of course, has created a big change in our workflow in that manner.

Yeah, and you kind of covered the other question that I wanted to ask. Basically, what does natural language querying actually unlock? and that was not possible before, so you kind of covered that. Curious to know there is real anxiety in analytics communities about AI making the analyst role obsolete So where do you actually stand on that Yeah, I think we have to really change the perception a little bit, right?

I mean, the analyst role is not supposed to be a gatekeeper of information or data, I think, in my mind, right? It should be more than that, right? It should be more like, you know, sort of thinking about strategic partnership, right? You know, so AI can serve that role, right?

Like AI can, or Gen AI tools and NLP tools, querying tools can serve, right? Like they connect the dots for the business stakeholders and they can query the data directly without the need of going through business analysts. But I think business analysts serve a very critical role because the need for like sort of like, you know, understanding the business contents, understanding like what is relevant question to ask, right? The AI, that will really kind of go into the decision making, right?

And also whatever AI is spitting out, right? It needs validation, right? So there are some of those aspects of the analyst role that will remain, right? Because, you know, you need that human in the loop, right?

You know, to ensure that the questions you were asking are the right questions. And secondly, the information that, like, you know, Gen AI engines are spitting out are the right sort of like, you know, contextual relevant information, right? And, you know, also guiding the business stakeholders to look how to ask the right questions, right? Those are the kinds of things that I think, you know, business analyst roles will sort of like still play very heavily, right?

So you stop being the gatekeeper of information, but you become more like a validator of information that AI is putting out and also then spend more time and, okay, well, scenario planning and kind of be that strategic advisor to the brand team. So you are sort of like in the middle, if you will, right? So AI interface is good to kind of answer some of the simple basic questions, but then more strategic thinking questions. I think, you know, you can take the support of AI and Gen AI, but then sort of like, you know, be that sort of interrogator, right?

as an interface between AI and business and also think about how do you really apply your business knowledge or that domain knowledge that you bring to the table as the insights hit the brand teams or hit the business units. That role is not going away, it's just evolving. I think in my mind. You have to understand that.

I think that's the great point you made. And that's very comforting also, knowing that there are so many intelligent people doing that role. They always have that kind of apprehension, which is important to be addressed. So, great point.

Moving on top of that, right? So, continuing on self-serve analytics capability only. How do you build a self-serve capability that empowers the commercial teams without creating the risk that they draw wrong conclusions from the data that they do not fully understand. Yeah, it's a good point.

I think a lot of it is like foundational, right? So capability, right? So if you think about one of the key things that has been coming up these days is, you know, how do you create that good semantic layer, right? Like for engineering engines to operate, right?

So one of the things that I think we are doing right in the middle of doing is, you know, we've created a data foundation, but then you really need a good semantic layer that really can, you know, you can ask the simple questions and understands the context and grabs that information. Right So like so AI is useful only when it is trusted Right So if you can build that trust with the stakeholders right in the interface that you providing them you know I think it will just fall apart right So to me, sort of creating those guardrails, making sure that the query responses are, you know, explainable, right?

It, you know, that whatever AI platform you're building, you know, it can document, right, like the way that reasoning, the reasoning and pulling the information, right? So it is, you know, it's documenting everything and it's kind of really giving you the explainable pieces, right? You know, as it's running the queries and that builds trust, right? For the stakeholders.

So it becomes really critical and that becomes a really part of your journey, right? Of self-analytics, right? If you are able to get to a point where, you know, the stakeholders can ask a question without the risk of them getting incorrect information, you know, that is number one. I think the worst thing that you can do is like, you know, you can create an engine, which is really great, and they can get the response, but then they make the wrong decision, right, based on that response.

So that is the worst thing that could happen. And that is where I think, you know, the partnership with the business unit leaders comes in, right, like stakeholders comes in, right, you actually are the custodian of, you know, the information and insights for a very big decision that they're trying to make, right. So, you know, think about, you know, creating the platform is great, but at the same time, thinking about, okay, well, you know, ensuring that there are card wheels and then the questions that they're asking are the right questions, right?

Those are the kinds of things that I think we need to start focusing more on, right? To build that trust, right, with the engine. And also make sure that you are creating that value, right, for stakeholders. Yeah, I'm sure.

I mean, I know you've been driving this within your company. So you must have gone through a lot of challenges also. in that. But the topic that I want to get into is what does it take to get a commercial organization to actually adopt self-serve analytics tools?

Even that service model which is existing in the industry, right, where analytics function is literally used like a service provider. So given that behavior change given that context the behavior change is often harder than the technology builds so i'm curious to know what does it take to get the commercial organization to actually adopt self-serve tools and how are you driving it in your company no it's a really good question and i don't think like you know we are there yet right so to me like we're in the middle of like this change management process right as we build our infrastructure and the self-serve tools you know to me it starts with um you know thinking about like how how stakeholders are consuming information right and what channels are they consuming information right whether it is you know so you know if you if you think about the best sort of like tools would be where you know whatever you build gets embedded in their workflows right so if you think about field teams right field teams absorb a lot of information from crm programs right like what crm systems that they are using right on day-to-day if you can build the workflow so that they get all the core insight directly within the CRM platform itself, it becomes their life easier.

They don't have to think about learning a new tool now. So making sure that the AI tools or whatever the Gen AI engine that you're building, it sort of like embeds within their workflow. You're not offering them a new tool or well, now you need to go there. So then it becomes like a little bit of a change management for them.

So how do you integrate in their daily workflows? How do they consume information now And then not trying to disrupt that too much I mean of course upgrades are fine but like how do you really not disrupt too much of their workflows on a day basis because if you are asking the stakeholders to learn a new tool that is challenging that is not their function right like they are there to kind of really you know make business decisions and drive the growth uh you know for a given therapy or whatever right uh to me like they are not there to learn new tools which which might be the analytics function right so so to me like how do you embed those workflows directly within the consumption layer that they are using right so that that becomes really critical and then i think that really reduces a lot of change management aspects right when they see that you know um you know thinking about like wherever they're consuming information it's all ai driven in the back end right uh it becomes much much easier i would say got it got it got it the other one i wanted to learn from you is how do you maintain the analytical standards and data governance when you are when you are actively pushing to broaden the access and reduce dependency on on central analytics function right in your case also you have multiple therapeutic areas you have you have a central analytics function so how do you maintain that analytical standard and data governance yeah that's a good point and i think there are certain things that you have to do centrally right um you you know, things like data governance and things like, you know, because the data governance principles would go across therapeutic areas.

It's not just one therapeutic area, right? But the access controls and the systems that you build should be embedded directly, right? So that it becomes a little bit of a, you know, you create that central, like, you know, governance, but at the same time, allowing the business units to operate independently, right? Because each other therapeutic areas have different business problems that need to solve, right?

Oncology might be solving, you know, a patient finding problem, right? Or the same thing as rare disease. Then there are, like, you know, things within immunology and ophthalmology that we are trying to kind of drive the growth for a given indication and so forth, right? So the business problems that we're trying to solve are different.

So we might need different sort of, like, custom analytics engine. But the idea is, like, the data governance and the access principles can be applied centrally, right? You know, one of the things that I think we can talk a little bit about is the Next Best Action program, right? So, you know, as you're thinking about the Next Best Action, the idea is like you create a technology foundation, right, which really allows you to kind of now become, you know, each brand can be onboarded to that capability.

And they can operate their own business rules very separately, right, across. So, you know, even then, the central role remains that the guardrails and access controls are applied centrally, but then like each business unit can customize their own. So that is where I think we need to have both aspects. It's more like a hybrid environment that you need to operate in.

At least that's what I've seen successfully. Thank you so much, Ashutosh. This was a wonderful conversation. What I'm taking away is that the insights and analytics function from reporting to proactive intelligence, from dashboards to sensor decisioning, and from insights delivery to action enablement.

that's that's the title for me thank you again for joining and sharing your perspective no brother this was really helpful thank you so much and hopefully audiences enjoy this conversation thank you so much thank you so much magazine yes yeah absolutely and for more interviews and insights please visit cvomagazine.