Building A Data Science Organization
Download as PPTX, PDF0 likes75 views
This document outlines how to build a successful data science organization. It discusses common uses of data science in businesses, including predictive analytics, customer segmentation, and fraud detection. An effective team structure includes a business analyst, machine learning expert, engineers, and quality control. Challenges include getting corporate buy-in, navigating data silos, and unrealistic expectations. To succeed, the team should identify early win projects, define a long-term goal, enhance data through purchases, and start with common starter projects like customer support analysis.
Building A Data Science Organization
- 2. How Businesses Use Data Science •Data Publication •Sentiment Analysis •Adaptive Marketing •Mass Customer Feedback Analysis •Personalized Analytics •Benchmarking •Predictive / Prescriptive Analytics (How to improve Credit ratings etc..) •Fraud Detection •Customer Identification and Segmentation •Risk Management •Advanced Business Intelligence •Competitive Analysis •Trends & Forecasting •Market Performance Internal Informational Services Business Operational Marketing Product Development
- 3. Team Structure DataScience Business Analyst Machine Leaning SME Engineering Quality Control • Business Analyst / Product Manager Strong understanding of the business unit, corporate vision and customer or corporate facing. Functions as an internal champion for a seldom understood team. Delivers and brings awareness of accomplishments and capabilities. Capable of defining a vision and roadmap for Data Science • Machine Learning SME Brings Data Analytics skills Strong Story Telling with Data & Technology skills Constantly updating technology skills in a bleeding edge, rapidly moving sector. • Engineering Go To Market, Productionize and Automate Eliminate the burn of reproducing output • Quality Control Outside of functionality, validate the reasoning and likelihood of results. Data fluent and customer centric Focused on does the data & story make sense
- 4. Challenges to Data Science Success • Corporate Maturity Data Science is a mid to long term investment Data Ownership must transform from siloed to centralized Output is frequently a Numeric KPI Expectations are often unknown and unrealistic, hard to make data sexy for average person Budgeting a department that is advanced & traditional research based requires CFO buy in • Data Science takes about 18 – 24 months for ROI Requires mapping out all available data in a company Span across Organizational Silos Navigate legal regulations, privacy policies, license agreements Identify possibilities and match with Corporate needs Develop vision and Roadmap Achieve buy, funding and resourcing Product Development requires existing Product Owners to co-own priority and delivery. Source additional 3rd party data and fill gaps
- 5. How to succeed • The first 3 – 6 months are critical to get introduced to the company • Data is political in most companies Identify projects that are not conflicting and have recognizable early wins Set teams up for immediate success and future buy in & collaboration • Define a moonshot Starter projects help build synergy Familiarize the team with the business and customers A moonshot brings focus and objective and drives the creation of a roadmap. • Buy Data Data enhancement is key to growing opportunity In B2C it provides adaptive behavior that increases interaction In B2B is provides Benchmarking that is impossible for your customer to build about themselves or competitors. • Common Starter Projects Customer Support issue identification / product deficiency / product usage What works in your product What doesn’t What has high value, requires investment to improve Sales Customer churn factors Usage or lack of Support calls - resolved / unresolved Market Identification Product / Customer Success Customer Opportunities