In the last week of Aug 2017, there were a couple of press releases and news articles that were quietly doing the rounds. There was a rumor that GE was contemplating selling "Predix" stakes, wanted trim down the "Digital Revenue" targets, and wanted to focus more on consolidation. This news might not have sent shockwaves among the Industrial IoT community, but for sure it wasn't positive news to my clients, who were in the same business as GE and had a considerable investment in the Industrial IoT journey. In fact, they decided to cut and freeze the budget until next year citing various reasons — among which was GE's recent reports in the media.
In hindsight, it is very easy to comment on "Why things didn't work," but it also provides us an opportunity to look into the problem and analyze what went wrong. Without getting into the postmortem analysis of Predix, let us try to see the bigger picture and understand why Digital Transformation is indeed a chasm. There is an insightful article from Forbes on "Why Manufacturing Companies fail in Digital Transformation". Surprisingly, Forbes states that 84% of the Digital Transformations in manufacturing firms are destined to fail, and Mckinsey says 65% of Digital Transformation initiatives do not take off past the initial pilot.
Let's take a look at some headlines.
Below is an interesting news article with some key points:
Some key problems were:
- Technical problems and delays with Predix
- Fragmented end customer and industry segments for Predix
- Building data centers of their own — a go-it-alone cloud strategy
- Lack of a clear business case for a measurable RoI
- Tradeoff between poor execution of traditional businesses and over-focusing on future innovations
Based on my knowledge and experience that I gained from a few key engagements, implementing Industry 4.0, and discussions with key customers, I have listed some of the reasons why Industrial IoT/Digital Transformation can fail and some key considerations that need to be taken care of.
Lack of Internal Alignment
While the drive and push for Digital Transformation comes from the top (management), that effect is not cascaded down with the same magnitude. Often, in a large organization, strategic initiatives related to IoT and Digital Transformation are spearheaded by the IT group.
Unfortunately, IT doesn't understand the domain and business problems that management is dealing with, so there is a huge gap between the groups. This creates a massive roadblock in the IoT transformation journey.
Also, many departments within the organization, like manufacturing or engineering, already have a pilot or small solutions, and the objectives of those solutions are not understood by the IT team. There is a radical shift in the focus areas of IT and the business group. IT focuses on technology, implementation, vendors, and maintenance while the core focus of the business group is strategy, business cases, processes, and people.
IT-OT integration is the key success factor for a successful, sustained Digital Transformation journey. Unless this marriage happens, Digital/IoT Transformation will be difficult.
Though top management initiates and leads IoT transformation, the change needs to begin with the front-line workers. Those above need to understand the challenges and problems faced by a quality inspector or a field service technician or a site engineer. Often times, the top management initiates a Digital Transformation journey by giving a keynote in an industry-wide event or announcing a significant investment in Digital to the board. Then, the implementation is handed over to the middle management. But middle management too doesn't really understand the problems that supervisors or front-line workers face. This situation is best explained by the "Ignorance Iceberg" depicted below
Also, most top management professions think that Digital Transformation begins with the customer. This is not entirely true. Digital Transformation needs to begin from within. It is critical to understand the stated and un-articulated needs of the shop floor supervisors, front-line workers, or the field service technicians.
This approach is pretty much aligned with the concepts of Design Thinking and Lean Methodology. This mindset involves looking into the problem with an empathetic, cultural, and even an emotional point of view. So, Digital Transformation needs to be first internally focused, then customer focused.
People Need Solutions to Their Problems, Not IoT/Digital/Cloud Technology
Companies need a solution to their problems. They don't need an IoT solution or a cloud solution. Often times, the Digital Transformation journey is taken with no clear goal or objective or business case in mimd. Industrial IoT is not a silver bullet or a one-size-fits-all kind of a solution. Many customers and consultants think, "Connecting a thing to the internet solves the problem," or, "Moving data to the cloud solves the problem."
Just because a device has the ability to connect to the internet, it doesn't mean we should. No customer wants to buy IoT — they want to buy a solution for the problem. The customer does not want to digitize or implement Industry 4.0 on a shop floor just because it is rated as one of the Top 10 technology trends by Gartner.
The key questions that need to be asked are, "Does implementing a remote monitoring solution yield an increase in service revenue or reduce downtime? "Does implementing a paperless plant result in increased productivity, improved cost savings, or a direct impact on my operations costs?"
These are the types of questions that need to be answered before taking the plunge into IoT. Sometimes, customer problems can be solved by a product innovation or a service innovation or a supply chain transformation. So, the key assessment that needs to be made is whether the proposed solution or the technology solves a business problem.
There is a radical shift in perspective when we look into problems and solutions from a Platform Thinking lens. The platform is a game changer and alters the business model beyond recognition. Each business problem modeled as business use case or a solution slowly begins to transform itself into a platform that has a 10x increase in value as compared with a solution approach.
As we begin to scale, platforms play a crucial role in determining the success metrics for Digital Transformation. It's very important to understand that platform ≠ cloud. A platform is a wrapper that is customized to match business and technical requirements to cater to a wide range of business problems in the ecosystem. Platforms enable collaborative multi-cloud environments that automate, manage, and optimize the use of cloud and analytics services at scale to solve a business problem.
The platform enables other business groups, suppliers, vendors, and system integrators to collaborate and create solutions that solve a business problem. The platform enables connecting and authenticating devices and sensors, deploying edge applications, aggregating data pipelines, integrating with existing sources of data, running analytics jobs and getting insights, and most importantly, providing access to internal/external developers, partners, and consumers of the solution through APIs that are capable of monetization. Roles such as Solution Architect are getting replaced by Platform Architects. Meanwhile, Product Managers are re-inventing themselves as Platform Managers.
Platforms: Building vs. Partnering
The decision to build or partner with your IoT platform is a make-or-break decision for an enterprise. This is where most of the investment gets sucked into a vacuum without knowing what is coming out of the other end. Large companies like GE have invested billions of dollars into building a platform like Predix, yet they are unable to reap benefits after 7 years, and course correction is still happening.
It can be very, very expensive to build your own platform, and it can take years to realize the RoI. The key factors that will affect this decision are the need for private cloud, how to onboard partners, how to create an ecosystem for developers, considerations around data security, how to enable launch platform updates and features on a timely basis, how to create a competitive edgen and what unique value proposition can be offered by your platform among the existing platform providers.
Unless there is an industry-wide or a conglomerate-wide need to build a platform, enterprises should keep away from this option. It is much easier and significantly less expensive to partner with an existing platform and build on top of it. Existing platforms already have a well-established developer base and a wide spectrum of existing services. They can leverage the strengths of the existing ecosystem, build a multi-cloud strategy, and get the best of services to create applications and solutions.
Most IoT solutions take off without due consideration about the monetization or the revenue model. While it is clear that products should move into the servitization mode, an important question remains unanswered. How do you design contracts that enable revenue from servitization? How does one convert the business use case into a financial revenue model? How do you determine which services to charge for and which not to?
Let's say it we are working with a case of anomaly detection for a gas turbine for a particular unplanned event. How do I charge my end consumers? Should I charge them for every anomaly I detect, or should I charge them for every minute that I am monitoring these gas turbines' operational parameters? Or should I simply have a subscription-based model or licensing fees for various classes of assets?
It all boils down to the business model and how to innovate it. The future competition is not really among products or services, it's among business models. Business model innovation will become the major source of competitive advantages. With advances in cryptocurrencies and flourishing APIs, there still needs to be a lot of work done on designing contracts that incorporate financial metrics and models that truly servitize products.
Data, as is said, is the new oil. It is, in fact, crude oil. Crude oil is almost useless in its current form and needs a lot of processing. Data is assumed to be always available, consistent, and reliable. But data from the real world is dirty, ugly, wicked, and extremely unpleasant to work with.
It turns out that raw data has a lot of inherent problems and causes nightmares to data analysts and data scientists. Often times, data scientists spend anywhere between 60-80% of their time just cleansing data. In industrial plant systems, where there are multiple protocols and multiple sources of data, things get a little messier.
How do you develop algorithms that get a good co-relation with future events? How do you link past data to existing events? How do you combine data from multiple sources? How do you develop algorithms that co-relate structured data with unstructured data? How do you treat time-series data?
In the case of IoT analytics (a.k.a. real-time analytics), you have to deal with key decisions at the edge. For instance, how much data do you collect and store? How often do you collect data? How much data can be processed at the edge? How do you ensure data consistency and quality?
All those considerations have a significant impact on the solution, architecture, and the platform design.
So, listed above are some of the challenges in the Digital Transformation journey. As and when we start building scalable and sustainable platforms and create many solutions on a platform, more challenges are bound to happen. So, it is absolutely essential to have the right platform and data management strategy, enable IT-OT integration, be inclusive of internal stakeholders, and drive Digital Transformation from within. Define a clear-cut business case and business model and, most importantly, develop a platform thinking approach.
Please note, these are purely my personal views and have no references to any client or industry or the company that I work for.