Why Most Data Projects Fail and How to Avoid It
summary
Claims that 85% of data projects fail and only 15% succeed backed by Gartner numbers lead to concrete prescriptions: prefer Snowflake/Databricks-aware pragmatism over vendor dogma, enforce a 5:1 engineer to scientist ratio, aim for a 10x ROI target, staff data engineering, data science and operations together, pick tech for the use case not the vendor, stage architecture complexity gradually, and measure success by getting models or pipelines into production with documented business value.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:50Technology is only part of the problem
- 2:34Right people at right ratios
- 6:49Business value of data projects
- 8:42When value is generated
- 11:32Where to execute data project
- 13:18Executing with a clear plan
- 17:10Inverting the problem
- 21:04Why copying architectures fails
- 24:41High vs low value teams
- 30:22Three roles for data value
- 31:05Q&A
related talks
29mBuilding successful data teams in the realities of 2023
22mBridging Business and Tech for Data Science Success
52mScaling Data Teams: Three Pitfalls and How To Avoid Them
32mWhy ML Projects Fail and How to Ensure Their Success
24mA Practical Approach To Unlock Value From Data and Analytics
57mWhy BI Projects Fail (And What You Can Do About It)
24mFrom Data to Decisions: Turning Enterprise AI into Real Business Outcomes
28mHow to manage data-related projects and not fail (too often)?