Data science has made immense progress, but companies are still stuck with the question: how do you use data science to deliver real value to the business? They hire dozens of data scientists and invest in state-of-the-art technology, but only a few have delivered ROI and business impact. In this episode, Wayne Eckerson and Alex Vayner discuss what organizations need to do for data science success.

Alex Vayner is a Partner and Americas Data & AI Practice Leader for PA Consulting Group, an innovation and transformation consultancy. Alex has spent his entire career in data & analytics, with his last five roles focused on building and running high-performance data science teams and capabilities in consulting and corporate environments. Before joining PA Consulting, Alex ran the NA Data Science & AI practice at Capgemini. He joined Capgemini from Equifax, where he served as VP, Global Data Innovation Leader, building a team responsible for pioneering disruptive data & analytics solutions for clients across all industries.

Key Takeaways:

• Apart from picking the right project, it's important to check data readiness and systems readiness.
• Some companies are successful in implementing smaller data science projects but are not able to industrialize the concept because of the unavailability of data and IT bottlenecks.
• Defining the problem is the most challenging part of the early stage POC.
• Business leaders have trouble coming up with outside-the-box questions can be used for data science projects.
• The more domain knowledge your data scientists have, the more successful you are going to be.
• The IT mentality of building something first and telling the business later decreases the proof-of-value.
• If you are planning to build new products using data science over a more extended period than hiring a team is recommended.
• However, if you only need to build one or two products than hiring a consultancy would be the right option.
• The build and buy decision depends on the price, control, and speed.
• Buying a solution brings speed, but the company loses control over pricing.
• If you feel that the insights could become a new revenue stream for you, then the buy option is not the best one.
• The benefits of cloud outnumber on-premise solutions.
• Instead of looking at the data, companies can look at models to organize and manage their data science data lakes.