Being Data-Driven is a Mindset Shift

Data-driven organizations are paving the way to more efficient, resilient and performance-rich futures.

This article will explore a deeper look into the what, why and how AWS is driving innovation with AI/ML and what that means for scaling data-driven organizations.

How to decipher this piece: This piece will be in a Q&A format to maintain clarity in breaking down what, why and how.

1. Browse categorized WHAT, WHY & HOW questions
2. Scroll down to Q&A
3. Glean insights from AWS specialists
4. Take a look at how globally known companies leveraging AI/ML with AWS
5. Learn what focus zones data-driven organizations should consider to scale long term

• WHAT? What does it mean to be a data-driven organization? What is Amazon’s flywheel model? What is AI vs ML? What are the key drivers for a data network effect? What precautions should be in place to prevent disruptions in data architecture? What services unlock CX (Customer Experience) improvements?

• WHY? Why should we consider data as a cornerstone for building people-centered technology? Why is a mindset shift key to fueling the flywheel at Amazon? Why is building a centralized data community crucial for improving data governance?

• HOW? How is AI/ML driving speed and efficiency in data-driven organizations today? How are consumers and producers impacted? How is ML being used in real time & which companies are using AWS?

What does it mean to be a data-driven organization?

• It is the recognition that data is an asset to create actionable insights to supercharge the customer experience. Adopting the data-driven mindset is an opportunity to influence each stakeholder’s narrative.

What is Amazon’s flywheel model?

• Let's take a look at Amazon's flywheel elements:

(Mindset + People + Process) x Technology

The concept is to keep the people, technology and the process at the center of all decision-making and using products, customer experience, feedback and data as learning models to experiment with to create innovation. Data-driven organizations recognize that it’s a process of working backwards from experience to create technology.

Why should we consider data as a cornerstone for building people-centered technology?

• Putting data at the center of decision-making enables organizations to: ➡ Make decisions, faster ➡ Respond better to the unexpected ➡ Create better customer experiences ➡ Uncover new opportunities ➡ Improve efficiency

Why is building a centralized data community crucial for improving data governance?

• It’s connected to how each sector of the community (producers, platform teams, consumers) extracts its business value.

➡ Producers: data sharing within teams to produce insights and KPIs, metadata and access considerations to retain owner governance

➡ Platform team: increases scalability and managing data compliance to move at the speed of business priorities

➡ Consumers: creators of new insights, data discoverers

Why is a mindset shift key to fueling the flywheel at Amazon?

• Mental models exist to drive and inform. Changing the mindset is key to fueling the flywheel because it involves placing value in who, what and why organizations create for and extracting insights from data directly is how the elements of the flywheel are enhanced.

How is AI/ML driving speed and efficiency in data-driven organizations today?

• Machine Learning requires a rich set of capabilities and data is the field for ML engines. As Tom Godden (AWS Enterprise Strategist) highlighted,

"When coupled with other implementations, AI has the power to change biz models…alone it cannot.”

• 3 key drivers for ML adoption: Relevance | Speed | Efficiency

What is AI vs ML?

• AI = Artificial Intelligence: involves the using computer science to learn from data models as humans learn from experiences.

• ML = Machine Learning: a subset of AI that involves using algorithms to learn and make predictions based on data.

• There are 3 main types of ML:

1. Supervised learning
   classification: predicts a category
   example: classifying a sock by color
   regression model: investigates the relationship between variables

2. Unsupervised learning
   clustering: groups by similarity
   example: sorting clothes by type
   association: clothes often worn together

3. Reinforcement learning
   environment based: do not need specific input/output pairs
   example: reacts to and learns from environment

What are the key drivers for a data network effect?

• ML is a tool to support business activations through customer needs to improve the CX. The cycle of The Data Network Effect thrives on 4 key drivers:

1. More data
2. Better algorithms
3. Better engagement
4. More, happier customers

How are consumers and producers impacted?

• By adopting the modern data-driven mindset, consumers can focus on engaging with products and producers can efficiently manage data retrieved from their experiences. This creates a culture of enhancing data discovery through increased visibility and trustworthiness, simplifying collaboration, working seamlessly with tools in context.

What services can unlock CX (Customer Experience) improvements (a non-exhaustive list)?

• Amazon Kendra (search)
• Amazon Lex (chatbot)
• QuickSight (business intelligence dashboards)
• AWS S3 (simple object storage)
• Amazon EBS (elastic block storage)
• Amazon Connect (cloud contact center)
• Amazon Transcribe (speech to text
• Amazon Textract (extracts data from any document)

What precautions should be in place to prevent disruptions in data architecture?

• To prevent disruptions, building resilient data structures that can manage the durability and performance of workloads. So how is this achieved? Through working with storage services like Amazon S3 and Amazon EBS.

Factors to consider for resilient structures:

➡ Resiliency: needs to match the workload that can manage the durability and performance of a workload

➡ Backups: back up volumes by taking point-in-time snapshots and storing them in S3

➡ Build your own redundancy: consider alternate routes for recovery, like Elastic Disaster Recovery

➡ Design for availability: no single point of failure should bring the system down (consider a load balancer)

➡ Monitoring, Testing & Recovery: simulate real-world conditions, create experiments to test

How is ML being used in real time & which companies are using AWS?

• The Washington Post - Textract, ECS, Docker, EC2
• NASCAR - Transcribe, S3 Glacier Instant Retrieval, SQS
• Condé Nast - Route 53, EC2, S3
• Fender - Lambda, DynamoDB, API Gateway, SageMaker
• Lyft - AWS Kinesis Streams, Auto Scaling, Redshift, EC2 Container Registry
• F1 - Lambda, SageMaker, DynamoDB API Gateway
• United Airlines - SageMaker, IoT Care
• Expedia - Aurora, Glue, CloudWatch
• Pinterest - S3
• BMW Group - Translate, Active Custom Translate, AWS Glue, AppSync
• Traeger - IoT Core, Alexa Voice Service, Lambda, S3
• Heineken - IoT GreenGrass

Data exists to be computed and stored, fed by efficient stage that can manage the data.
Ryan Sayre

It’s an exciting time to learn and engage with AI, and with the release of several machine learning services like OpenAI and search engines such as You.com, the globe is being taken by an AI-powered storm. As Godden highlighted in his AWS Innovate session 'Building a Smarter Organization: Powered by ML',

"When coupled with other implementations, AI has the power to change business models...alone it cannot."

AWS is paving the way to a more cohesive, resilient and performance-rich future, and the companies listed above are living proof of the power cloud computing has created. The more producers, platform teams and consumers engage as data-driven communities, the better the experience can improve for all.

➡ Think Big ➡ Work Backwards ➡ Design & Build