Industry has successfully adopted a “pull” approach to solving many of its problems. First popularized in Toyota’s production system, this approach is remarkably effective at supply chain management and assembly line optimization. Now pervasive in all aspects of project management, “pulling” leads to quantifiable results.
So why is machine learning (ML) and AI being “pushed”, if not shoved, forward at industrial companies?
The technology sector is leveraging machine learning in remarkable ways. By processing massive amounts of consumer data, these firms are predicting our buying behavior or movie-going tastes with extraordinary precision. Why can’t this technology work equally as well at industrial companies?
External consultancies can advocate this line of thinking; This ends up “pushing” machine learning within industrial firms. The problem begins by setting expectations that simply cannot be met. The industrial sector grew up with data analytics. The analysis of production data is in our DNA; we want to predict process behavior as accurately as we can. But, a naive adoption of methods and principles from the technology sector leads to disappointment — at massive scale: 65% of industrial firms report not realizing value from their ML and AI investments.
At Fero we recommend that our customers follow a 3-step “pull” approach instead.
- Define a real use case (define the “pull”)
- Empower your team use ML (first-time “tug”)
- Adopt ML to adapt to the future (continuous “follow”)
Define a use case (“Pull”)
Pick a real use case, not one that matches what the technology sector is doing. You can minimize your raw material costs, increase production yield, minimize emissions, and much more. What are the most pressing matters on your mind? Focus on use cases that will align all stakeholders with little effort.
Empower your team to use ML (“Tug”)
Your colleagues want to know one thing: “what can I do with ML technology”? Slide decks and unsustainable software written by external consultants will only delay the inevitable. Industrial experts have the best understanding of their processes — with the right ML software, they can learn how to “tug” at use cases and figure out the root cause of critical issues.
Adopt ML to adapt (“Follow”)
With actionable next steps, you can “follow on” and leverage ML to its fullest potential. ML can perform real-time optimizations to help operators make optimal decisions during production. Deploying ML on one use case gets others thinking about how ML can actually make a difference in their respective areas. This leads to more domain experts “pulling” in real use cases — the cycle then begins anew.
This approach yields real returns — Fero customers regularly achieve a ROI of over 250% in their first deployments.
If you would like to learn more about how a “pull” approach could lead to real results, please feel free to ask for a demo. We would be happy to learn about your challenges and how Fero may be of help.
