The hype around digital transformation among manufacturers has led them to stockpile data for problem solving and analytic decision making. While manufacturers continue to accumulate masses of data, most have yet to turn this asset into a tool for proactively managing operations.
Most of this data isn’t suitable, or a fit for the most critical problems manufacturers are looking to address. Unfortunately, even when the data is applicable to a given problem, it’s often not ready, or in a condition that makes it usable for analysis.
This data fitness and readiness ultimately determines the success of manufacturing analytics efforts. While manufacturers may be well-equipped in one or the other, both are essential to identify and operationalize insights from data.
We have found that most manufacturers lack an understanding of their data’s fitness and readiness. It’s critical to understand these concepts as they determine which projects are feasible and why timelines for different initiatives can vary widely.
So what is Data Fitness?
Data fitness means having the right data to support a defined use case.
We’ve worked with numerous manufacturers that have made significant investments in collecting data that they ultimately determine is not applicable to the use cases that deliver the biggest impact. It’s similar to having a well-stocked kitchen pantry that doesn’t have the ingredients for a specific recipe.
It may sound obvious, but many manufacturers don’t spend enough time understanding how their priority use cases align with the data they have in their kitchen pantry prior to beginning initiatives.
What is Data Readiness?
Data readiness is based on the data’s accessibility and usability. Continuing with our kitchen metaphor, it’s akin to not only having the correct ingredients on hand but ensuring they are in a condition (chopped, pureed, etc) that the recipe calls for.
For manufacturing analytics efforts, typical readiness challenges associated with accessibility include:
- Data is stuck in a PLC that is not on the network
- Remote manufacturing facilities without reliable network accessibility that prohibits them from streaming production data in real or near real-time
- Legacy production software that doesn’t allow for data export
Typical data format readiness issues that impact usability include:
- Timestamps that are not synchronized across different data sources involved in production
- Proprietary data formats that can’t be easily ingested or integrated with data from other sources
The highest possible level of data readiness describes data sources that are in well-structured, portable formats, and that are accessible via a network.
Simply having a well-established data payload is not enough.
Most manufacturers have little to no understanding of the readiness of their data for integration and analysis. Why? Because they haven’t tried to use data in a scalable manner.
At Sight Machine, we have worked with numerous manufacturers who thought they were effectively collecting and storing data. However, when it came time to extract and use the data for a project, they discovered the way it was being formatted during collection rendered the entire batch unusable.
We’ve even seen manufacturers who assume that capturing and storing data in spreadsheets will be sufficient for analysis. One-off data dumps might be good enough to support an individual analysis but cannot support a scalable, automated solution.
In order to build analytics that can be used to manage operations, manufacturers need to build a process for continually ensuring data is ready.
Having the right ingredients in the right condition is the key to digital manufacturing success
As we’ve mentioned, there is a surprising lack of awareness of how data fitness and readiness can impact project timelines and budgets among manufacturers.
CIOs and IT leaders need to understand their organization’s data fitness and readiness before beginning initiatives. Understanding these data attributes enables them to effectively prioritize solutions (‘What can I do now that creates value?’) and determine what needs to be done in the future to get value (“What do I need to address to deliver the most important use cases?”).
Simply having data payloads, establishing data lakes, and integrating systems for data pass-through is not enough. Understanding data fitness and readiness is critical to becoming a data-driven manufacturer.
In future blogs, we’ll dive into specific recommendations for understanding and improving data readiness. For some additional thoughts on data strategies that support manufacturing analytics efforts check out the blog by one of Sight Machine’s data scientist, Katarina Struckmann.