#243 Mastering DataOps and MLOps: Building a Strong Foundation for Success and Future Growth

At Data Futurology’s OpsWorld conference in March, a panel of experts came together to discuss the importance of getting measurements, processes and methodologies right to drive DataOps and MLOps across the organisation.

The panel consisted of Katherine Fowler, Head of Business Transformation at L’Occitane Australia, Amar Poddatooru, Head of Data and Technology at Australian Ethical, and Emyr James, Head of Data at Resolution Life and moderating the discussion was Andrew Aho, Regional Director, Data Platforms at InterSystems. It became a far-reaching discussion that started with methods to define and measure the ROI of data and analytics initiatives and how to get those projects off the ground. The discussion moved on to overhyped technologies in the data space, and then looked forward to what is on the horizon for the years ahead.

As the panel discussed, there is a lot of interest among consumers in some innovative technologies, including ChatGPT. This is in turn driving a lot of interest at the executive level at rolling out solutions that use these tools. However, without the right foundations in place, and without proper concern for the privacy and regulatory risks associated with these tools, they will cause the data team more headaches than they’re worth.

This panel discussion is essential for understanding how to structure a foundation for data success, be disciplined in deploying the available resources across the data team, gain executive buy-in, and then steadily build the practice up.

Enjoy the show! 

Thank you to our sponsor, Talent Insights Group!

Join us for our next events: Advancing AI and Data Engineering Sydney (5-7 September) and OpsWorld: Deploying Data & ML Products (Melbourne, 24-25 October): https://www.datafuturology.com/events 

Join our Slack Community: https://join.slack.com/t/datafuturologycircle/shared_invite/zt-z19cq4eq-ET6O49o2uySgvQWjM6a5ng

“I know that across industries, data quality is one of the biggest problems that's been there for years and it's still not going away. I wish I could turn data quality into a dollar value and convince executives how important it is to have good data quality.”

Emyr James, Head of Data at Resolution Life

WHAT WE DISCUSSED

2:07: Felipe introduces the Measurements Thought Leaders panel and moderator, Andrew Aho.

3:48: How do you define and measure data and analytics ROI?

7:21: A discussion on metrics that help get data initiatives off the ground.

9:41: How a data leader needs to focus on the data platform, and articulate both the “big picture” view and the details.

12:35: As more organisations adopt ops, processes and methodologies, what challenges might people anticipate arising, and how can those be addressed?

17:24: What can data professionals do to help solve the change management challenge?

18:34: What are the challenges and impact of upcoming “silver bullet” technologies like ChatGPT?

20:16: What is currently overhyped in the data space (and why)?

24:03: What can we as data scientists do to ensure that we’re looking at the right risks and drawing accurate conclusions on what is right for the business?

26:13: If the goal is to focus on data science, how can we also keep experimentation and creativity going?

29:49: How do you estimate the value of change to get executive buy-in?

31:18: What upcoming developments and trends will emerge over the next five to ten years?

EPISODE HIGHLIGHTS

• Katherine: “I'm looking to understand what it costs us to do business this way right now. So then when we put in place an improvement initiative, I can actually prove a dollar value ROI to the people who speak in dollar values.”

• Katherine: “We want to understand how many people are moving to the report we've just delivered, and then we ask what that improvement means for them. Depending on who it is, it may actually be a difference between cost reduction and revenue generation. For example, the difference between sales and marketing and their adoption of a report versus someone in the executive suite.”

• Amar: “When I worked in a retail organisation, it was very hard to show the business the value of data and analytics. One of the organisation's strategic objectives was to free up $500 million. How can we achieve that? One of the things we discussed was getting better at inventory management. Retailers have 15, 20 distribution centres. They have lots of stock sitting in there, so if you optimise your logistics, if you optimise your supply chain, and if you optimise your stock management, you can actually bring that down and free up the capital. In the end, building that analytical model to reduce the stock within the warehouses then allowed the company to go and invest in the other markets.”

• Amar: "It's not always about the top line. It's also about the bottom line as well. Bringing operational efficiencies. The last thing we wanted was a scenario where the customer comes into the store, but the stock is not on the shelf. It would have meant that yes, you have achieved your 500 million reduction target, but then if you are also creating a bad experience for the customer when they're walking in.”

• Emyr: “I know that across industries, data quality is one of the biggest problems that's been there for years and it's still not going away. I wish I could turn data quality into a dollar value and convince executives how important it is to have good data quality.”

• Amar: “Resistance to change is something I see, and is probably one of the biggest challenges. What I would say is to address that issue is to engage people early, get their buy-in quickly, but that doesn't mean engaging every data user who is going to be part of this journey. Rather, it is really important to get the sponsor’s buy-in early.”

• Amar: “Integration is a challenge, and people have lots of choices. There are too many silver bullets out there. So I would say don't choose too many platforms. I know companies with five or six different integration platforms. I would say just choose one or two to solve the key problems and then stick to them.”

• Amar: “We have evolved so much. The data scientists and the data teams are now taking on more accountability. If you look at traditionally what happened 15, 20 years ago, you used to have a DBA, you used to have a BA, you used to have a testing team, infrastructure team. They used to do all their own different things, which they needed to do to achieve the business outcome. But now we’ve got this data scientist or a data engineer - call them a super engineer because they got all the keys to the kingdom - and then they can do anything they want. But at the same time, they have to start learning all these new skills.”

• Emyr: “If you can show to execs and prove to them how you get value from data, then you’ll get their buy-in a lot earlier. I still think that execs somewhat think that technologists will be technologists and will want to play with tools.”

• Katherine: “As more of these machine learning and AI tools come to market and get hype from the general public - and that I think is Chat GPT is a particular example of this - then quite often you get pressure from higher and higher up to implement these tools, regardless of whether there's a business case and a clear strategic direction for the implementation of the tool.”

• Katherine: “A challenge when the general public hypes a technology is that when you are looking at the cost-benefit of implementing a particular tool in the business space, you really need to be much more mindful of things like data security and especially if you are bound by requirements such as GDPR.”

• Katherine: “Quite a number of organisations start with a very small team when they're very early in terms of maturity. This is a difficult conversation to have. Obviously, you’ll need more people to get to the goal of implementing AI and similar solutions, but if you can't prove from the beginning with that tangible ROI, you won't reach that level of resourcing that you need to ultimately prove success at a much larger scale.”

• Katherine: “I would not consider experiments in the early stage of a data analytics program. You need to have proven the worth of the work that's being done to be able to say to the relevant people, ‘All right, give us space, give us resourcing, et cetera,’ to be able to look at what we could do next.”

---

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. And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us.