After you have a plan for gathering and analyzing your manufacturing data, there’s 3 types of insights you can take advantage of: Descriptive, Prescriptive, and Predictive. If organized and accessible data sounds far-fetched, check out our resource on fostering a data-driven culture.
Descriptive insights start with clean data you can rely on. Reliable data allows you to make logical decisions, which is where the Prescriptive functionality comes into play. Lastly, once you have a big bucket of successes and historical data, you can use AI to predict those successes in the future in a Predictive data model. We’ll dig deeper into each of these.
A Descriptive data model is the starting point for fully exploiting your data. That means having information that is accurate and up-to-date. It should be helpful and easy for you to get the information you need. You should be able to rely on this data to make informed decisions. There are many vendors out there that can help you get up and running with Descriptive data. Gartner Research, for example, has a lot of material already out there about how you can convert your facility to gather more data.
This all ties into the most fundamental parts of your business. Running a Lean facility will have you looking at your processes carefully, and it’s worth evaluating if there’s a good way to tie those processes back into your software. Having up to date inventory in your ERP and analytics is valuable for looking at inventory right now, and it’s even more valuable looking at trends over time.
For the most part, these are all things you know and work in every day. How much do things cost now versus then, how long does it take to make them, how are they selling? We’re not talking about re-inventing the wheel here, just ensuring that you are answering those questions accurately all of the time and can refer back to that information in 6 months, 9 months, or even further back.
After you have good, reliable Descriptive data, you can start to make decisions with it. Some of those decisions might become routine to the point that you can automate their output. For example:
“The cost of raw materials has gone down, and we will increase output because we can maintain a higher profit margin.”
These types of insights are based on criteria specific to your business. Your manufacturing manager, for example, would be able to work with the Analytics program to see patterns, and then assign basic steps to take when certain criteria are met. That’s helpful for a few reasons:
- It institutionalizes data and their insights in a platform multiple people can access
- It provides helpful tips that new employees can quickly use
- Processes can be developed around identifying and implementing profitable improvements based on data
Soon these steps will be partially taken over by Artificial Intelligence, which will mean your continuous improvement and management teams will spend even less time in the mundane processes. The last piece of the Data-Driven Manufacturing model is the predictive component, which we’ll cover next.
The final step in a Data-Driven Manufacturing model is using machine learning to make predictive insights about the future.
The key to making this work is that you have to have plenty of clean historical data to base the predictive model on. This is fundamental to Artificial Intelligence.
AI and machine learning have a lot of hype around them right now, but it is relevant to this discussion about data. You need deep, accurate, historical data to do statistical analysis and predict future trends.
That’s all AI is: really good statistical analysis. If your data is accurate and reliable, you can use a machine learning process to make predictions about how your manufacturing environment will change. That insight can then be used to make adjustments in real time that drive value back to your business.
Creating a model like this is not out of the question. In fact, innovative supply chain companies are already doing this around the world. Right now, Germany has almost 70% AI adoption in their manufacturing, while companies like John Deere in the US have acquired machine learning technology companies to kickstart their AI implementation.
The easiest thing you can do right now is start thinking about how your data is a resource, and what you need to do to properly exploit it. For any questions along the way, don’t hesitate to give us a call at 405-421-0644. We’ll have one of our engineers get back to you within a business day or less.