Episode 191 Part 2: Product design and development with Sarah Dods, formerly Head of Advanced Analytics at AGL

In this second podcast with the visionary data scientist, entrepreneur and business leader, Sarah Dods, we dive into the details around product development and the role of data science.

Across her career, from NICTA and CSIRO, through to AGL, Telstra Health, RMIT University and Gerson Lehrman, Sarah has had a long history in bringing innovation to market.

One thing she says is that business leaders should not lose sight of how data science needs to work. It’s a team sport, in the same way that building a car requires more than someone to design and build the engine.

“So, what do you need to make a solution supportable, sustainable and safe?” Sarah asks, before going on to note that another challenge that data science teams need to be cognizant of is that data science models fail silently. The application could be bringing garbage in, and pushing garbage out, and it will happily keep working, producing outputs that will no longer mean what you want them to. So teams need to think about your feedback loop.

This is where agile comes from, in encouraging an iterative process in which the MVP is produced as quickly as possible, and then iterated on indefinitely as opportunities for ongoing development arise.

Stay tuned for some deep insights from Sarah about the opportunities for data science to drive next-level product innovation.

Enjoy the episode!

Thanks to our sponsor, Talent Insights Group

Discover what the Data Futurology Community is up to...Join here 

“Data science is a team sport. The same way that you build a car. It requires more than someone to design and build the engine, you need the rest of the car, you need an upholsterer, you need that diversity in the team to understand the different perspectives.”

- Sarah Dods, formerly Head of Advanced Analytics at AGL

WHAT WE DISCUSSED

0:00 Introduction
01:38 Any examples that come to mind in terms of showing the discrepancy between initially having a focus on building the engine and then describing the problem as getting from A to B, and focusing on how the car drives?

12:17 We think of data science, almost as projects by default, what do you think about the difference between the two? What should we think about it?

18:01 What are you passionate about in this space? And what are you frustrated about in the space?

19:20 Where do you think people fall down or can fall down in the inclusion component?

25:58 I was keen to ask you about your frustrations in our market space, in our industry. What do you think could be better, should be better? Anything that comes to mind in that space?

EPISODE HIGHLIGHTS

• If you're in a startup, where you've got some idea around how to do something inside the corporate world, it's actually almost the other way around. You've got a problem, then you need to find the right solution for that problem and approach it from the other end and work backwards.

• Having that longer-term perspective, but with an urgency to find out the answers quickly creates the space for a really successful product to be able to be developed.

• Try to understand what the world looks like through different people's eyes, and what's the piece that different people bring to the table. Make sure that you listen to them. The inclusion part is really important.

• The thing I tell myself around those when they happen is it means that the fact that somebody was willing to bring that up means you successfully created psychological safety. Because they feel they can bring it up and then they trust that you will actually do something about it.

• One of my frustrations is that there's the success rate of machine learning and getting it into operations is so low. That's a steal that 80 to 90% of data science slash machine learning initiatives don't ever make it to production.

• The scientific method is different than developing hype, you don't develop hypotheses and test them in it. You just don't it's a different way of approaching things. And yet in data science, it's absolutely fundamental to what we do.

• I guess I come from a world where you find your customer and get it right or you die. And when you're looking at corporate analytics, that immediacy of risk is not there to the same extent and then just a different way of approaching things. 

---

At Data Futurology, we are always working to bring you use cases, new approaches and everything related to the most relevant topics in data science to help you get the most value out of these technologies! Check out our upcoming events for more amazing content.