WELCOME
DATA SCIENCE STRATEGY Are we ready for it? Asure / DOMO 3
DATA SCIENCE STRATEGY – ASURE’S EXPERIENCE Bruce Harris Ulises Gonzalez-Guerra Director IT and Enterprise Business Applications, Pricing Specialist / AWS Cost Optimization Project Manager Guest Lecturer Graduate School of Accounting University of Texas at Austin Asure Asure 4
TOPICS WE WILL COVER What does the Data Science Strategy Consulting Project consist of • Obstacles Asure faced on its way to Data Science Readiness • Building the Asure Analytical Model • Lessons provided and applied working with DOMO Data Science Consultants • 5
KEY TAKEAWAYS • Don’t let data constraints define your Data Science Vision Develop your Conceptual Model • Design and create your Master Data Set • • Short Term Wins Long Term Vision 6
DATA SCIENCE STRATEGY CONSULTING PROJECT • The Readiness – Developing the capacity to execute a Data Science Strategy The Modeling – Leveraging the tools you have created during readiness to build • and test a Machine Learning Model • The Production – Going live with your machine learning model 7
Building the Asure Model
APPLICATION ECOSYSTEM Customer Support Tax CPM Expenses CRM Integrations Marketing Gamification of Travel Expenses Quote to Cash Analytics ERP
GOALS Introduce Domo and analytics to a wider audience • Use analytics to drive decision making and create revenue opportunities • Customer Health, Customer 360 – know our customer better • Improve forecast accuracy • Churn reduction • Prospect identification and conversion • 10
OBSTACLES No analytics solution (before Domo) • Disparate, disconnected data • Lack of In-depth knowledge of steps needed to create analytical models • 11
SCENARIO Work with the Domo Data Science team to learn foundations of analytics and • models Create Master Data Set in Domo • Work with Domo Data Science team to turn master data set into working • analytical model 12
OUTCOMES In-depth knowledge of our customer • Master Data Set that can be used by the whole company – no more looking for • data wondering if it is correct Prospect Monetization • In-house knowledge of Data Science best practices • 13
FUTURE Use feedback loop to further fine tune analytical model • Find more uses for Master Data Set • Continue to improve our knowledge of Data Science best practices • 14
FOUNDATION {PHILOSOPHY} Customer Customer Renewals CaPDB 360 Health Master Data Set Analytics Driven Root Cause Analysis Master Data Management No Connected Scalable Cloud Code{Configuration}
COMPONENTS OF A DATA SCIENCE MODEL Outcome – Behavior you want to predict (churn, revenue, prospect • conversion): dependent variable Predictors – what affects the outcome: independent variables (AR balance, • payment history, product count) Controls – other factors that are related to the outcome (industry, geography, • credit score, number of employees) Errors – affect the outcome but cannot be accounted for in the model •
BUILDING THE ASURE MASTER DATA SET
ANALYTICAL UNIT OF MEASUREMENT What unit of analysis will all your variables be recorded across? • Think rows, not columns • Necessary to define which format your data will revolve around •
Domo Master Data Set Customers Accounts 370+ Data Points Financial Risk/Credit Risk/Payment Risk # Employees SF Account ID # Locations NS Customer ID Address Primary Key Industry Codes Sales Orders/ Annual Revenue Invoices Opportunities NS SO ID SF OPP ID Macro Economic Data Domo Domo Connector Connector Employment Data Master Data Set c Salesforce Account NetSuite Customer Salesforce Opportunity NetSuite Sales c Order/Invoice c Transactional Domo Dun & Bradstreet Data Connector Data Asure Transactional Data Cases Domo Macro Economic Data Connector Zendesk Cases
Build the Customer Profile Master Data Set Salesforce Account NetSuite Customer Salesforce Opportunity Customer Profile NetSuite Sales Order/Invoice Dun & Bradstreet Data Asure Transactional Data 1 Record Per Customer Macro Economic Data Zendesk Cases Per Month Unit of Analysis
Domo Data Science Models Generate Customer Churn & Revenue Forecast Customer Churn Customer Profile Revenue Forecast
Prospect Conversion Prospect data cannot be loaded Into HubSpot for 370+ Data Points Campaigns due to Financial Risk/Credit restrictions. Risk/Payment Risk # Employees # Locations Create Prospect Address Account Records in Industry Codes Marketing Salesforce Annual Revenue Prospect Campaign Accounts Using Domo Generated Customer Profile, Send D&B Request for Prospects Current process is Reactive Matching Our Master Data Set (pulls credit report after Standard Profile Salesforce Account deal is closed. Proposed NetSuite Customer process is Proactive (credit Salesforce Opportunity score is known before NetSuite Sales Order/Invoice prospect is contacted), Dun & Bradstreet Data limiting churn risk due to Asure Transactional Data financial condition. Macro Economic Data Zendesk Cases
Process Prospects in Prospect Analytical Model Using Domo Data Science Prospect Accounts Opportunities Feedback Customer Master Data Set Prospect Master Data Set Loop Salesforce Account Salesforce Account NetSuite Customer Dun & Bradstreet Data Domo Salesforce Opportunity Connector NetSuite Sales Order/Invoice Send Closed/Won Dun & Bradstreet Data Results to Domo Asure Transactional Data Macro Economic Data Zendesk Cases Generate Prospect Domo Conversion % Writeback
ROI – TARGETED PROSPECT CONVERSION * Typical prospect conversion rates for SaaS Software companies are 0.5% to 7.5% Number of Targeted Prospects Conversion Rate Prospects Converted ROI Requested 25,000 0.1% 25 -45% 25,000 0.5% 125 173% 25,000 1% 250 446% 25,000 2% 500 993% 25,000 5% 1,250 2,632% 50,000 0.1% 50 3% 50,000 0.5% 250 417% 50,000 1% 500 934% 50,000 2% 1,000 1,967% 50,000 5% 2,500 5,068%
ROI – CHURN MITIGATION Churn Reduction ROI 1% -76% 2% -52% 5% 20% 10% 140% 12% 189% 15% 261% 20% 381% Prospect Conversion Analytics would also contribute to Churn reduction by limiting prospects to companies with strong financial scores. 25
THE DATA IS EVERYWHERE Where is the data? • What data do we need to accomplish a specific goal? • What process do we need to implement to start generating the necessary data? • 26
CHURN – CONCEPTUAL MODEL • Credit Score • Billing Terms • Number of Checks Processed • Customer Life - Bill Last Processed Check When they are Up for Price Increase • • Number of Payrolls Processed (Contract Renewal) • Industry System Up-time/Downtime • • Number of Employees at Company - feature Classic vs Web Customers (Platform Type) • • engineering (turn employee count into Last Upgrade Version • factor) Time since last Up-sell • • Date/Time Pricing • Number of Products/Product Mix • Geography • Macro-variable (Employment, GDP) • Customer Service Contacts • Time to live • Customer Payment History • Customer Contacts (weak maybe) -time Payment Terms since last support case submitted • 27
DOMO SCRIPTS AUTOMATICALLY CREATE DETAILED DATA PROFILE Summary Statistics 28
DOMO SCRIPTS AUTOMATICALLY CREATE DETAILED DATA PROFILE Correlation Matrix 29
DOMO SCRIPTS AUTOMATICALLY CREATE DETAILED DATA PROFILE Histogram Data Profile 30
KEY TAKEAWAYS • Don’t let data constraints define your Data Science Vision Develop your Conceptual Model • Design and create your Master Data Set • • Short Term Wins Long Term Vision 31
AND ONE MORE THING… 32
33
THANK YOU
Recommend
More recommend
