A presentation at Databeers VLC in in Valencia, Spain by Alan Byrne
Machine Learning. The Product Perspective. Alan Byrne @thealanbyrne www.alanb.ie
Leveraging ML to achieve Business Goals... Enhanced Customer Experience ● ● ● Personalisation and Customisation Discovery & Inspiration Contextual intelligence More Efficient Internal Processes Creation of New Products and Services ● ● ● ● Product Availability and Fulfillment Customer Service Automated Decision Making Leverage existing data and insights in new ways
Amazon X-Ray Real time content insights
Pinterest Find Similar Items. Shop the Look. How should I style this?
Examples from Google Smart Search Smart Replies Smart Notifications
Disney Book Ears Speech recognition adds sound effects at certain points throughout the story
OK, so Machine Learning is amazing, let’s do it! STOP! ● ML is not a magic wand for your business or product! ● ML is a solution - you need to first define the problem: what is it you’re trying to achieve? ● ML may not be the right solution for a lot of problems. ● With ML it is very tempting to dive in and start building models… Could you solve the problem without it?
The Nature of Data Products You don’t go into data product development knowing exactly what you’re going to build or how you’re going to build it or whether it is even possible...
The discovery journey... Valuable Usable Ethical Feasible Viable Gather evidence!
But how…? Smart Product Development Agile Data Science Organisational Structure Team Composition & Culture Principles and Practices
Smart Product Development... Talk with your customers Prioritise ruthlessly Tackle risks & assumptions Establish baseline value first Focus on the problem Be data informed
Agile Data Science... Iterate, Iterate, Iterate Ship Intermediate Output Collaborate & Communicate Experiment more than Implement Keep it Simple Listen to the data
<7 mins…?
Machine learning provides many opportunities for enhancing the customer experience of existing digital products and services or, enabling the creation of whole new product experiences. But the challenges of building and launching these products and features require special considerations to maximize their chances for success. This talk will examine the approach to building data products and features touching on Agile Data Science, Rapid Prototyping and Lean Experimentation.
for free. You
can too.