Interview with Andy Teoh Part 1

Thank you, Anna! My name is Andy, and right now I am the marketing and commercial obsolete in Organon and Organon is a global health company dedicated to making a world of difference for women, their families and the communities they care for. So prior to joining Organon, I worked in a few tech companies, consulting companies and startups. Myself, I’m a data scientist by train and I had the privilege and the chance to be one of the leaders in commercial in Organon, and I hope that I’ll be able to spend time with you guys to be able to answer questions that will benefit each and every one of you.

How has the pandemic forever changed how businesses and what would be the value of data science in the next couple of years?

Thanks for the question! It is really a great privilege to be able to speak at World Data Summit.
I think that this prolonged pandemic situation is the tipping point for digital transformation in business. I’m not talking just about digitising every single asset or just taking the business from offline to online. But, with digital, you actually open a lot of opportunities. And suddenly in the past 12 months, as you can see, there are many, many players jump into the bandwagon of digital. And what I would say is that companies that can be very quick and nimble at innovating will tend to be more successful. So with digital connectivity and IoT, we know that the volume of data growth is unprecedented. So what this pandemic really brought forth is an explosion in the growth of the data. And this would mean that the value of data or the ability to synthesise and extract valuable insights from data is going to be so much more important in the next couple of years. And, you know, what you need is that for businesses, they are going to be able to understand their customers better than their competitors in order to be in that business. And this is going to need a lot more data to help them to innovate, given incremental innovation in the products or services or even come up with entirely new products and services or other possibilities to your existing business. So this pandemic has really changed the way we think and the way business looks and data is going to be so much more important.

We all know and have read about the power and value of data, but why still so many companies out there struggle to realise the value of data in their organisation?

Yes, I think most of us, the commercial leaders, are very convinced that this polling data and data will be the differentiator for many businesses. But as you have rightly asked this question, why is this so difficult? Because the natural thing is that if you talk about harnessing unlocking the value in data. It is not a simple task. It is a very, very tough job. And why is it so tough is because you need to do well in science and by science, I mean that you need to have a sound architecture in the enterprise, you have to develop a good pipeline for your data when you are building the model to extract insights from the data you need to use the right mathematical model, you need to choose the right model in order for you to get the insights from the data. And then also the art piece which is how you communicate those insights to the business, to the commercial stakeholders? Because at the end of the day, you want insights to be actionable and you want your analysis to be relevant to the business. And very naturally, when we talk about science, there is a disconnect between both of them. And if I were to say that in the typical organization, the commercial unit and IT unit usually talks different languages, and in order really for us to harness the value in data, we really need to get the whole process from the size piece to the art piece to be right in order for us to really utilize the value of data in the business.

Now, I just want to give very specific examples and experience that I had when I was working in a company to try to improve data centricity. Everybody starts to use data. So what happened is that when I was in the organization, I’m being tasked to do a centralized system that can offer a recommendation to the marketer, to the salesperson, what is the next best action to take? Now, the one I learned from this particular example is that when we are working on it, we have a very good team to gather the data together with you to model. It just sounds right in the science piece. The pipeline is robust. The model, it is one of the best model that we have built. But at the end of the day, when we give our recommendations to the marketers, to the sales team, they start to question us. “You have built a very good recommendation. You recommend us to do action A and action B.” People like to be exposed to more of such content or services. But, you know, in the eyes of the business is that those are not really the critical piece. As in how they want to use the insights in their marketing. So when we talk to people in marketing from the science side, we usually set up a target verbal and then we put everything else in. Just to see what affects sales. But, you know, actually, it’s more than that. It is not just as simple as that you just pick a few variables and you just tell them that “hey, you know, I’m going to recommend you to take this path”. But we have to understand that in this process we have to onboard our sales and marketing team very early from the start so that we put in the right assumptions and we don’t build our model and share the insights based on the assumptions that we make. So this is one particular incident.

And if I were to draw another example. Just imagine a data scientist working in Walmart. He or she, if he works alone, is very good at science, gets all the data, working very hard to find out what are the insights.

And you may spend a lot of time getting the insight, so what happens if the Data Scientist runs to the CEO’s office and say that “hey, Mr. CEO, I have run a market basket analysis and I found out that bread and butter are two commonly purchased items in a basket. And what the CEO would like to say: “Thank you so much for your hard work. I don’t need to run to market basket analysis to know that these particular two items go together”. 

And if I were to change the context. So if the Data Scientist from Walmart this time goes to the CEO and says: “hey, you know, beer and diaper go hand in hand. It’s a commonly purchased item in a basket”. I think this is a very good insight and that will strike like “OK, tell me more about it”. That will actually trigger interest. And if you can deep dive a little bit down and say, “oh, usually the profile of the customer who purchases diapers get the other debts”. In the US where usually the Mom will ask the Dad to just go and buy a diaper and when they are buying a diaper, usually, they will just grab a loan.

So then that will be a very interesting insight. The insights are also something that we need to work together with the marketers. Do you want to put beer and diapers side by side so they increase the sales? But I would say that in terms of visual merchandising, it is not really visually appealing to put a diaper beside a beer. Another action that you can do is put the beer and the diaper at both ends of the shop. And then when the Dad kicks up a beer or a diaper, he will walk all the way to the other end and pick up the diaper. So in a way that you might encourage the Dad to buy more items in-between.

And so as I begin to talk about all these things, if you look at harnessing the value of data in the businesses, there are so many things to take care of. It is beyond just setting up a good architectural done. It is beyond just like building up a good model and say I’m done. We need to carry out to really shed those insights for companies to take action.

In the current climate, what now are the TOP 5 criteria for a successful data strategy?

I would say that the criteria actually depends on the maturity of data in the company. It is very important for us to really harness to get the data to be right. So No.1 is you need the science to be correct, meaning that we need to have the right kind of architecture to fit the company’s purpose. And we also need to think about what are some of the capabilities, the science capabilities that we need to insource, what are some of the things that we need to outsource. It may work for you, but it may not work for other companies in some of the industries.

No.2 is the organisational structure or structure design and the mix of people that we put in, whether we want to put the technical team to be a centralized team, a decentralized team, or some sort of a hybrid. You have a centralized team and then you also have some teams that are in the market as well. We’re going to talk about the mix, whether you want to have more engineers or you want to have more scientists or analysts and where you put your scientists and especially engineers. So I think that it requires really careful design.

And then No.3 is how good are we at bridging the gap between commercial and IT? I think a few years ago McKinsey published a report that there is a need for the analytics translator. So I think the rise of such roles is really talking about we need people to be good at talking to languages to bridge that gap.

No.4 I would say is the talent competition. At the end of the day, the success of the company really depends on the type of talent that we are able to attract to nurture and keep. So I think it is really critical that we get the right kind of talent, skill sets, and capabilities to be in the organization for us to be very successful.

And then the last one, I would say, is that we need to think about how can the organization embrace agile? Gone are the days when I approach the IT department to answer my business question. I think right now a lot of people would know that in order for us to really adequately address a business question, we can run the first second cut of the analysis to shed some insights. But at the end of the day, you also need to adopt a mindset of testing. I always say that the value of data is great. We are able to get a lot of insights, but it doesn’t tell you the absolute truth. Tt gives you a reasonable doubt and you have to go and test out those reasonable doubts. Once you have those insights, you also need to test out what is the right kind of actions and then the ability to capture a better outcome as you have designed. 

So these are the five things, you know, in my opinion, would be very important for an organization to be successful.