Why ML Projects Fail and How to Ensure Their Success
summary
Recent Gartner estimates put ML project failures at 60-85% failure rate and the panel gives concrete mitigations: treat projects like an investment portfolio with t-shirt sized risk ratings, require business metrics and mid‑manager buy‑in, map desired decisions to available data, and don’t proceed until data engineering is ready. Use short pilots—sometimes a four-week fail-fast check—and aim for 6–12 month ROI; one client claims analytics work could have saved building a new refinery.
AI-written summary · about
chapters (8)
- 0:00Introduction and panel welcome
- 1:24Defining ML project failure rates
- 8:19Key causes of ML project failures
- 15:43Evaluating ML project portfolios
- 20:59Failing fast in heavy industries
- 26:14Convincing skeptical customers
- 28:44People as the greatest risk in ML
- 32:00Closing remarks and thanks
related talks
31mWhy Most Data Projects Fail and How to Avoid It
27m10 Things That Can Go Wrong with ML Projects
24mFoundations for ML at Scale: Where to Start in Your Organization
42mMasterclass: The right path to ML
27mKeys to Building Machine Learning Systems
32mWhy Most Data Projects Fail and How to Avoid It
41mChallenges in incorporating ML models
59mWhy Great Models Fail: Lessons From 9 Years of Deploying ML Models