How to Build a Manufacturing Analytics Platform

Developing a manufacturing analytics platform is no small feat. Trust us – we’ve done it! 

Some manufacturers may decide to develop their own manufacturing analytics software to predict defects and automate root cause analysis, just like LinePulse does. However, these internal development efforts are not always successful. 

Why Some Internal Manufacturing Analytics Development Projects Fail

At Acerta, we’ve had six years of dedicated focus on developing an analytics platform for manufacturing. We’ve been introduced to internal development efforts of various types and at various stages. We commonly see these projects fail for a few reasons: 

  • Poor understanding of the problem to be solved and how success will be measured 
  • Underestimating the time and effort required to build and maintain the platform 
  • Inexperienced or over-burdened project leadership 
  • Not investing in the necessary resources to fully develop the platform 
  • Not considering critical aspects of platform design such as UX design, scalability, speed, and training 

It is only natural that a manufacturer might see their analytics platform project go off the rails for one of these reasons — and it isn’t something to be ashamed of. You simply haven’t had the experience to teach you the ins and outs of software development. Manufacturing is your core competency and where you excel. Just imagine a tech company trying to enter the automotive manufacturing space. In fact, Apple tried just that and ended up spending billions of dollars before scrapping the project. 

While we suggest that a product like LinePulse may offer you a better return on investment compared to building your own solution, we’re happy to share what it takes to build one based on our experience. 

What You Need to Build a Successful Manufacturing Analytics Platform

A Clear Vision for the Project

Before developing an analytics platform, it is important to have a clear, long-term vision about what problem needs to be solved, and what the outcome should look like. 

To justify dedicating resources to an internal build, you must be certain that your unique domain knowledge enables you to create a tailored solution that will result in an increased return on investment versus an off-the-shelf solution. Do your research and do the math. 

Some other questions to consider are: 

  • What are the primary business goals the analytics platform aims to achieve? 
  • How will the platform improve decision-making and operational efficiency? 
  • How will success be measured? 
  • Who are the primary users of the platform? 
  • What are the plans for future updates and feature enhancements? 
  • How will the platform adapt to changing business needs and technological advancements? 
  • How will the introduction of the platform impact existing workflows and processes? 
  • How much will platform development cost? 
  • How long will platform development take? 
  • What happens if the project is unsuccessful?

This project vision must be fully thought-out and periodically re-evaluated. 

Project Leadership That Understands Both Manufacturing and Software

Leading an analytics platform development project is not for the faint of heart. It requires experience and understanding of the software development process, the manufacturing environment, and the challenges that come with building industrial applications.   

We recommend that the team be composed of at least two individuals so that focus can be divided between the strategic vision and the day-to-day implementation. 

Product Manager   

A skilled product manager with a proven track record of successful software development makes an excellent leader for developing manufacturing analytics. They understand how to bridge the gap between the product vision and what is technically possible. A product manager must balance a wide variety of considerations to plan what the platform will do, what it will look like, and how it will be used.  

The product manager will also be responsible for stakeholder engagement, reporting on KPIs around cost savings and efficiency improvements, supporting adoption and implementation plans, and more. 

Development Project Manager 

A development project manager is essential to keep the development efforts on track and within budget. Although their focus will be more on the day-to-day activities of development, they will work hand-in-hand with the product manager to drive the development process.  

A development project manager must have a thorough understanding of the software development cycle to be able to accurately estimate timelines and assess progress, which is more cyclical than that of other industries such as construction or manufacturing. 

Resource allocation is also part of their job. They can help identify skill gaps in the team structure and estimate the size of the team necessary for the size of the project. 

A Large Enough Analytics Development Team

Every software development team will look slightly different, depending on the size and nature of the platform being developed. It is important to ensure that the team is large enough to employ skilled specialists in each area necessary. Gaps in the team’s knowledge can cause the project to run into serious problems that no one can predict. This can result in major rework and delays. 

These are the essential roles necessary for developing manufacturing analytics: 

  • Machine learning engineers 
  • Data scientists 
  • Front-end developers 
  • Back-end developers 
  • Site reliability engineers 
  • Application engineers 
  • Technical writers 
  • Software QA specialists 
  • Security specialists 
  • UX designers 

It is common for manufacturing corporations to already employ IT staff members at the plant or corporate level that may have some knowledge or ability in the above-mentioned areas. We caution against leaning on these employees as full-time members of the analytics platform development project, as they are unlikely to have the capacity to take on an entire development project in addition to their regular duties. If other projects compete for their attention, they may cause the entire development process to stall. These IT staff members may be quite skilled in supporting and maintaining data and IT systems but lack the experience in designing and building something new from the ground up.

Manufacturing analytics platform development is a long process that requires a coordinated effort from dedicated employees to be successful. The smaller the team, the greater the risk to the project should one of the team members leave. Project leadership should anticipate these risks and plan their project documentation accordingly. 

An Understanding of How to Cross the IT/OT Chasm

The analytics development team must be prepared to take on the challenge of gathering, collecting, accessing, and normalizing industrial data. In software development, this process is typically referred to as ETL (Extract, Transform, Load). The ETL process in industrial environments is more complex and nuanced than in other industries due to the variety of industrial communication protocols such as Modbus, OPC (OLE for Process Control), PROFIBUS, Ethernet/IP, and MQTT. 

Some of the challenges in crossing the IT/OT chasm may include: 

  • Implementing best practices for data security 
  • Collecting data through different industrial communication protocols that have different structures and communication methods 
  • Barriers to compatibility that existing software data sources may have 
  • Managing system resources to process large volumes of manufacturing data without latency issues or data loss
  • Creating the correct data infrastructure for a data pipeline that is in line with the desired analytical methods 

A Plan for Long-Term Usability and Adoption of the Platform

Dealing with the data, development, and the technical side of generating analytics is often the sole focus of inexperienced analytics development teams. Creating a tool that is accurate, effective, and functions properly is extremely important. But if that tool is challenging to access, slow to use, or not intuitive in the way it displays and shares information, it will deter the end users from using it. 

No matter how powerful an analytics platform is, it will not achieve the desired business results unless it is actually used. From the early stages of development, it must be designed to be user-friendly, fast, accessible, and designed in a way that the analytical insights the platform creates are understandable enough to be put into action. 

There must be a plan for who will use the analytics platform, and how existing workflows will be altered once it is fully functional. Ensuring that the platform is adopted in the long term involves the creation of technical documentation and training materials, interactive user training sessions, developing a system for technical support, and handling change requests and bug fixes. 

A Plan for Updates and Future Development

Technology and industry standards are constantly evolving. We don’t use the same file formats or programs we did 10 years ago. Analytics software must be designed to be compatible with other changing software and hardware in the plant for as long as it can. When file formats, browsers, or operating systems change, how will the platform be updated to be compatible? And is the version 1.0 design considering these upcoming changes? 

Other factors can affect the lifespan of software, such as changing security considerations as new ways to hack and compromise data are invented. Future regulatory requirements or customer quality traceability could also factor into how an analytics platform can be used. 

The point is, creating a software platform is not just a “one and done” endeavor. It requires strategic planning and ongoing maintenance efforts to be successful. 

In Summary

If you are still deciding whether or not to build or buy, make sure you do your research. Get a full understanding of what it takes to undertake a project like this.

From our (admittedly biased) perspective, there are many reasons that purchasing an analytics platform leads to a higher return on investment than trying to build one internally. Usually, building your own platform is quite costly, takes a lot of valuable time, and can be a distraction from focusing on what you do best manufacturing.

If you’ve decided to leave the platform building to the experts, book a meeting with us. 

Share on social:

Automate root cause analysis and predict defects in real time

How is that possible?