Digitizing Subject Matter Expertise Part 3: Implementing Digital Knowledge Strategies

Note: This is the third blog of a three-part series that examines human expertise as part of digital transformation. Digitizing subject matter expertise is critical for companies that want to implement prescriptive, agile operational processes, such as maintenance. The term for doing so is called digital knowledge. Digital knowledge is the discipline of encoding human expertise and data from across silos, digitizing these insights, and then driving them into relevant decision flows. Analytics are key tools for digital knowledge, though many companies have yet to use them in this way. If you missed them, read Part 1 and Part 2.

Implementing Digital Knowledge Strategies

To remain competitive and thrive, industrial companies need to integrate human expertise into digital transformation. Doing so means overcoming the barriers inherent in tribal and stranded knowledge and capturing that expertise. Digital knowledge is the means for doing so.

Digital knowledge is the discipline of encoding human expertise and data from across silos, digitizing these insights, and then driving them into relevant decision flows. This enables the subject matter experts to make better and faster decisions that improve operations.

Implementing a digital knowledge strategy requires a specific, though relatively straightforward, methodology. This combines a knowledge-centric orientation, a range of analytics techniques, artificial intelligence (AI), and technology. Companies wishing to do so should:

  • Begin with the business problem and relevant subject matter experts, not data. Lack of a business problem-centric orientation is often the undoing of many digital transformation projects. Companies too focused on data become overwhelmed by the volume and complexity. Reactively, digital transformation initiatives become a series of complicated IT projects, as leadership deems that group most capable of managing data issues. When companies begin digital transformation by first defining the business problem, the expertise and decision flows needed to solve it becomes central to the task. As they know the business best, SMEs drive the knowledge creation process, adding the necessary context and decision flows for solving the problem. By removing the gap between problem identification and the knowledge needed to solve it, companies accelerate the speed of digital transformation.
  • Let the data format and sources guide the analytic techniques. When it comes to implementing knowledge-centric digital transformation, both simple and complex techniques can be applied. What is important is that the technique be suited to the data, rather than applying a one-size-fits-all approach. These data are likely to be a mix of structured and unstructured formats, including work logs, applications, event reports, images, emails, manuals, historians, the internet, etc. A range of techniques may be needed, from decision trees to cognitive analysis, for example.
  • Speed continual improvement by removing data science skill barriers for SMEs. Once engaged in digital transformation, SMEs will want to continually innovate to improve and simplify aspects of the business that impact their role. However, they lack critical advanced data science skills. Eliminate that gap with tools designed to assist knowledge-centric analysis, such as semantic search, natural language processing, drag-and-drop data mashups and blending, guided and auto-generated models, and model libraries.
  • Amplify knowledge to create new value. As decision flows become digitized and operationalized, the potential exists for connecting them to drive additional value. Secondary and tertiary interdependencies can be identified and leveraged, providing a more holistic, informed view of the business. The insights provided by the connected decision flows can create new value. For example, the flows can be pushed out as role-based recommendations to any affected personnel, including engineers, finance, and field workers.

As with any business transformation, achieving timely, measurable results determines success. Some of the world’s largest industrial companies are using digital knowledge strategies to drive significant business benefits. Many are realizing results in half the time of prior efforts, all without having to bring in an army of consultants.

Comments

  1. Dear Mike, I totally agree with your observations and conclusions, but I was hoping for more specific pointers to practical solutions. In my company (Pragma in South Africa, asset management solution and service providers), and in my role as Strategic Consultant, we regularly identify this challenge (and opportunity) at most of our clients, where critical tacit knowledge is not well exploited, and rapidly dwindling as the SMEs go off in their directions. Another specific occurrence that I find, is where OEM and highly specialist individuals are engaged during turn-key or EPC projects, and then “refrain” from parting with their highly valuable knowledge once the project is handed over. The innocent owner/operator then enters an extended O&M life-cycle stage, and is soon faced with a sustainability risk when maintenance planning and delivery becomes inevitable. This risk could be largely mitigated with good and efficient access to captured subject matter expertise, i.e digitized knowledge as you refer to it. A case in point here in SA, is the increasing number of new renewable energy plants, that are now typically going out of their initial warranty periods, and becomes the owner/operators responsibility for the next two decades.
    With reference to your specific recommendations above: the first bullet provides a good pointer, for an efficient approach to build momentum, but then the next two bullets, although surely true, are very open ended conclusions. A more practical example or case study, would be highly valuable.
    Please contact me, if you are willing to provide more insights.

  2. This was a great series Mike! Our company, Mobideo, bridges the gap of the information silos and encourages knowledge transfer. Let’s connect and chat.

  3. I personally agree. I see two ways human expertise is driven into workflow decisions:
    1.) Reliability and maintenance engineers use analytics apps where SME expertise is encoded into fault models and operational models that help them determine the condition and performance of these equipment like pumps and heat exchangers so they can plan maintenance and other activities accordingly.
    2.) Real humans reviewing analytics like equipment condition and performance reports before it is sent to end users to take action, as a connected service, and support in case of doubt.

    I personally agree that subject matter experts need to be involved in discovering the needs and defining the requirements, to ensure the tools become useful to their teams. Without the involvement of maintenance and reliability engineers, those teams may end up with general purpose analytics tools meant for data scientists, not at all suitable for the equipment performance and condition monitoring they need to do. The SMEs must evaluate the datasheets and other documentation for analytics tools and select purpose-built tools that meet their needs for plain text reporting etc.

    The choice of analytics technique is important. Machine learning requires a learning period to learn failures. However, because equipment failures are expensive and far apart, a learning period is not practical. Right?

    I personally agree that being too focused on the data, especially the location of the data, may start the project off in the wrong direction. For instance, creating another middleware platform layer to aggregate data which is already aggregated in the historian in one more place is overreaching. Instead tap into the data from wherever it is. Focus on how to collect the missing data.

    Adding another layer of data aggregation platform can become a very complicated project like you say

    I agree that different analytic techniques will be used. For instance, the real-time analytics for maintenance, reliability, and production used by operations departments uses different techniques from the transactional analytics used for business intelligence (BI) by other departments. The BI tools will likely use the output from the operations analytics for operations-wide analytics.

    I personally agree that analytics tools must be easy to use because the real end users of the analytics such as reliability and maintenance engineers etc. are not data scientists. Therefore using general purpose analytics tools is not practical. These personnel need purpose-built tools for equipment condition and performance monitoring analytics.

    I personally agree that software and apps must be role-based. The reliability engineers requires data different from that of an energy manager. Thus they need different tools, purpose-built for their task, not general purpose.

    Key to on-time on-budget roll-out of digital transformation is to use building blocks (sensors, analytics, security, mobility, and connected services) designed to work with each other and your system; a complete digital ecosystem, not just a “platform”. Indeed the building blocks shall be platform agnostic such that it can work with your existing system. The analytics apps shall be readymade to avoid custom programming. Use robust and deterministic analytics, embedded with codified process automation and equipment knowledge. Role-based software and apps purpose-built for the task, not general purpose.

    Learn how other plants deploy easy to use analytics
    https://www.linkedin.com/pulse/real-end-users-analytics-data-scientists-jonas-berge