IoT

Have the Right Partner Ecosystem To Avoid ‘Pilot Purgatory’ With IoT Projects

Many companies get stuck in pilot purgatory when it comes to implementing IoT projects. In this article, discover how to avoid this state of limbo.

July 21, 2021

The gap between an organization’s aspirations for an IoT project and the realities of moving from proof of concept to full production is known as “pilot purgatory” by some pros. Gathering trusted partners and managing the organization’s analytical maturity can help you avoid this state of limbo, say Bobby Shkolnikov, global principal in the IoT division, and Bryan Saunders, head of industry consulting, IoT division, SAS.

Industrial Internet of Things (IIoT) initiatives are ubiquitous within the manufacturing sector. Unfortunately, organizations tend to get stuck in “pilot purgatory” rather than being able to move swiftly from a proof of concept (PoC) to pilot to full-scale implementation. Organizations often aspire to rapidly achieve lofty levels of intelligence with their use cases but are disappointed when they fall short. Avoiding that disappointment means sidestepping common pitfalls such as not designing for scale, not considering the deployment or management of various AI/ML models over time, and not having the right partner ecosystem in place. 

Scale Out: Evolving from PoC, to pilot, To Full-Scale Implementation

Recognizing value from a variety of IIoT use cases hinges on an organization’s ability to transition from one maturity level to the next. It often starts with choosing the right technology partners and systems integrators to help you prove a concept and help you push to the pilot phase, where benefits accrue. And they’re the expanded teammates who can help you gracefully move to programmatic level changes that make a lasting impact on your organization. 

Also Read: 4 Must Haves of an IoT Connectivity Solution for Enterprises

There are many “random acts of digital” that exist under the manufacturing and supply chain umbrella, and few of them make it to full implementation. One of the challenges we have seen regarding converting the PoC/pilot to a full-scale implementation (and later to a managed service) is less about the core technology components and more about the support, documentation, stability, ongoing maintenance and durability of the solution.

Architectural decisions centered on proving out value in the earlier phases (PoC / PoV and pilot) skew towards reducing cost components. As a result, a lot of open source products are selected.  While open source is a viable solution, it can create a chasm for scaling out toward a larger implementation. It often leads to incurring technical debt. In the field of analytics, for example, those challenges often come from managing and deploying models rather than creating them. 

Start by choosing enterprise-grade partners committed to making an investment to prove the value and can achieve scale without incurring technical debt, such as tailoring a portion of the solution to their individual specifications. Additionally, partners that have the capability to drive governance and deployment, especially around model ops when it comes to deploying analytics in an IoT space, can help offset initial costs and pave the way for more successful future scaling.  

Scale Up

Expediting how quickly we move a use case along the maturity curve is known as scaling up. The goal is to layer in new functionality to capture greater value. For production quality use cases, for example, the evolution trajectory follows these steps:

1. Statistical process control

Manufacturing equipment is often instrumented to provide visibility into the behavior of key performance indicators that are critical to successful production execution. For example, a food processing machine may measure the temperature in a drying process to ensure optimal moisture content in finished goods (i.e., a sandwich cracker). Analyzing that measure against defined control limits helps ensure that the process remains within quality specifications.

Most PoC/Pilots start here with a focus on maintaining consistent “inputs” to the process by eliminating special cause variation that would lead to costly scrap and rework operations.

Also Read: All You Need To Know About eUICC Enabled SIM and Why It’s Important for a Greener Planet

2. Predictive quality 

As organizations begin to understand the critical variables for their process, they typically transition to advanced analytics to give them insight when the process drifts out of control. As opposed to reacting once a critical parameter falls outside of its operating limits, they begin to rely on predictive models to flag anomalous behavior. For example, if the drying temperature for making a sandwich cracker remains at X, but the humidity drops below Y, and the feed rate slows to Z, there is an increased chance of a defective product. These early warning signals allow for process adjustments to reduce the risk of a defective product.

Few PoCs/Pilots are here as large amounts of contextualized data are needed, and tools/technical capital is not always available to execute on this. However, this is a good stopping point for many initiatives as the art of the possible is displayed, and a business case can be made to justify additional investment.  

3. Autonomous control 

An emerging area in the field of manufacturing uses advanced technology to support visual inspection and assessment of subassembly and finished good quality definitions. This allows for increased testing coverage and greater accuracy by eliminating human bias. With real-time quality inspections, organizations unlock the potential to drive autonomous and closed-loop control systems that can react in real-time to changing process dynamics.

This level of maturity is rare. Few PoCs, pilots, or full-scale implementations make it to this level of integration due to the technology and process investments required.

Also Read: How IoT and RTLS Together Are Powering Businesses Everywhere

Takeaway 

Moving up this maturity curve requires artificial intelligence (AI)/machine learning (ML).  Building/managing/deploying the AI/ML models takes time and is difficult to do. However, there is a tremendous value that can be achieved. Start small and focus on business outcomes to determine whether the next step up the maturity curve justifies the investment at your current stage of transformation. 

Scale Out and Scale Up Together 

Value is generated by scaling out and scaling up individually. But there is synergy in synchronizing how we scale out/up to maximize value, minimize investment risk, and create a self-funding mechanism to drive a digital transformation and avoid pilot purgatory. This can be done by bringing in the right ecosystem of partners that span service providers with deep domain knowledge, software partners that bring enterprise-grade solutions to accelerate time to value, and hardware vendors to enable the end-to-end vision.

Did you find this article helpful? Tell us what you think on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We’d be thrilled to hear from you.
Bobby Shkolnikov
Bobby Shkolnikov

Principal - Global IoT Strategy & Commercialization, SAS

Bobby Shkolnikov is a Global Principal in the IoT division at SAS. He focuses on building partnership ecosystems and has helped organizations in the high-tech, manufacturing, and life science sectors.
Bryan Saunders
Bryan Saunders

Head of Industry Consulting, IoT, SAS

Bryan Saunders is the Head of Industry Consulting the IoT division at SAS. He provides expertise to drive advanced analytic solutions across industry in support of IoT.
Take me to Community
Do you still have questions? Head over to the Spiceworks Community to find answers.