Manufacturing Insights: Beyond Democratization of Data

Imagine what could be accomplished if everyone has access to the data they need and knows what to do with it.

The democratization of manufacturing data won't unlock the value of the Industry 4.0 era. Despite all the industry talk about it, democratization of data isn't enough. You need to focus on the democratization of insights.

You need to get actionable insights based on the real-time data into the hands of people who can use them. These insights inform and support business objectives, such as fewer shutdowns, improved health and safety outcomes, higher productivity, better throughput, and increased profit margin.

But will your people trust theses insights? Every business will experience this somewhat differently, but there are challenges we consistently observe, and tried and true tactics that work.

The amount of data isn't the only issue.

There's so much data to manage, and so many different types of data are required to paint the whole picture. Manufacturers need the infrastructure to support the seamless integration of data from a variety of disparate sources: data of different types and in different formats, structured differently or not at all.

To make the most of your data, you need to be able to ingest all that disparate data and and make it useful and available to analytics, automation, and machine learning (ML) solutions. You also need to know what questions to answer, and what datasets to select.

This is a whole-of-business task.

Democratization of data is all about making digital information accessible to everyone across the business that needs it, whether they do or do not have technical data skills. Employees shouldn't need to request help from the team, coding, or new tools every time someone needs data access. It's time to enable the sharing of data across multiple parts of the business and put an end to the gatekeeping.

Democratization of insights takes that to a new level, involving the right mix of ML, analytics, automation, custom algorithms and more to draw useful insights out of the right datasets. It involves mixing data with suitable external datasets. And, crucially, it means feeding those insights out to the people who need them, when they need them. You can't make it IT's responsibility to figure out how to sort all of that out. They can't read the minds of everyone who needs data. Include the people who will use the insights, so you know what questions you need the data to answer.

Recognize that different roles may require different views of the same data, to rapidly comprehend, contextualize and action the information and insights from this data.

It's no good if people don't trust the data.

Involving a cross section of your people early in this process is a best practice. That includes people who have always based decisions on gut feel and experience. You need ML to make the kinds of decisions these subject matter experts would make if they could combine that experience with all the data and processing power of your advanced analytics solutions.

Invite your team members to participate in discussions with the data engineers and data scientists around how the analytics work, how ML algorithms are trained, or AI systems developed, so that trust in the data and the process can be strengthened. The initial results of a predictive algorithm may not be very accurate. However, models can be continually improved with the support from the team. Then, when you start democratizing insights, they will recognize those insights as having been developed with their input. That's powerful.

The interface and user experience matter.

If your colleagues have been using the same tools and processes for a long time, they are likely to be resistant to new ones even if you promise them better information.

Invite team members to contribute to the design of the user interfaces. Consider solutions that work within your existing ecosystem, integrate with familiar tools, and even present the insights via existing dashboards or trusted user interfaces. Most solution providers want to sell you on a single-pane-of-glass approach. And, while that's the ideal outcome, that pane of glass can be your current system with new, modernized solutions in the background.

Ease of use goes a long way towards user adoption and return on investment.

Solve specific challenges.

To achieve true business value, you need to quantify your technology investment against real business objectives. Here are two timely examples:

• Tackle management of hygiene within manufacturing operations by leveraging data gathered via industrial IoT devices and applying advanced analytics. Monitor this data in real time and feed predictive, actionable insights to your personnel.
• Reduce the impacts of supply chain disruption on your production flow. If your team members are fed insights based on a real-time assessment of production and assembly processes, they can dynamically reschedule processes or substitute components. This kind of flexibility is mission critical when component supply can be suddenly interrupted by border closures or other shipping delays.

With the world still in a state of uncertainty, it's more important than ever to be data driven. To do so, you must be able to make your data available, reliable, and presented as actionable insights. Quality insights lead to improved decision-making. Democratization of insights gives a business significant competitive advantage when all team members embrace the knowledge generated.

[Originally Published in AMT (Austrian Manufacturing Technology) Magazine; October 2021, Original Link]

Check out more great stories on Insights.