How data prep on AWS empowers Autodesk customer analytics
- 2. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. How data prep onAWS empowers Autodesk customer analytics John Gardner D E M 2 9 - S Senior Business Intelligence Analyst Autodesk
- 3. Agenda Introduction • Autodesk • Market Intelligence Challenge/Use Case • Developing a 360 view of a sub-set of customers to drive marketing and sales strategy • Current process and its issues Solution • How our team has leveragedTrifacta & Snowflake on top ofAWS to streamline our previously manual processes Results • Time/effort reduction • Increased stakeholder collaboration • Deeper analysis
- 6. SampleProject Lifecycle Extract /Compi le Manipulate Publish RefreshPhase Descripti on Issue
- 7. Segmenting customers based on marketing & salespotential Sample Request Create a prioritized list of current construction customers that do not own our construction processing tool (BIM 360) Challenge • To create proper prioritization, we need to create a full 360 degree view of our construction customer base • Data used to track customer financial metrics is housed in a separate data warehouse than account and product ownership data • These two data sets also have different unique identifiers mapping needs to be done using a separate mapping table with >1m rows • Separate data sources w/ various BI access points + large data sets + inability to automate = lengthy, unstable process
- 9. Internal Enterprise Sources External Marketing and Market Research Sources
- 11. Transform, Automate Organize Thenewprocess isasimple,streamlined,and dynamic Powered by RefreshPublish
- 13. Manipulate
- 14. Results
- 15. Thank you! © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
- 16. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Editor's Notes
- #9: Extract: Require account profile data and financial data which are only accessible via separate BI tools. Thus the info must be downloaded into excel Datasets ultilize different unique IDs so a separate mapping table must be created to union the data. This mapping table is built off of a larger mapping table and must be created using an IN function for every ID (sometimes over 100k), creating a slow and inefficient process.