Why So Many IoT Projects Are Failing In Fortune 500 Companies

Why So Many IoT Projects Are Failing In Fortune 500 Companies

Posted by Machine Sense on

For last 4 years, I talked to or collaborated with a large number of internal IoT teams of Fortune 500 companies. Most repeated theme I found -they are struggling to grapple with complexity of IoT against the backdrop of high expectation of their CEO/board. Hardly any IoT project has been deployed in large scale. Most of them are in POC phase. Everyone is feeling the burn with swaths of open issues in large scale implementation. It will be worthwhile to discuss those in details so that everyone can benefit from the discussion.

First of all, let us discuss why most of the IoT projects do not go beyond POC phase:

1. Commercial Viability - where is ROI ( Return on investment) in any IoT project ? Unless, the project is about monitoring critical equipment, sensor cost and cost of managing IoT system has to be extremely low. Sensor cost is coming down but cost of managing IoT network & system is at its infancy or has not even started. Thousands of ways wireless network, sensors, API, electronics, gateways can go down defeating any benefit of IoT what so ever. To keep the IoT system running is a BIG cost. That's where most of the IoT projects are losing viability unless a huge investment is made to automate IoT management and support.

2. Adaptability of new IoT system by clients : B2B is a different ballgame for adoption of new technology and lifestyle. Smart Phone can change lifestyle of all the generations within a decade. But Industrial 3.0 which started around 1970s made very slow progress and even to date majority of Industrial machines run on analog I/O. IoT systems call for a different kind of operational management and not many of the operational managers are willing to make that change. You can send the SMS/alerts but if the operational manager feels that to manage the operation, the SMS/alerts must be integrated with existing MES/ERP system, project will be at toss because cost of those integration will far exceed any benefit specially when no one knows the exact benefit with certainty.

3. Security is not standardized: Every IoT system is highly vulnerable and easy to break into. Reason is easy. To manage IoT device, every IoT system allows OTA or over the top patching to update those devices. One can secure it in many possible ways- Asymmetric encryption, identity management, two way authentication and many more. Still every thing can be breached fairly easily. DLT ( Distributed Ledger Technology) is trying to change that by recording all the transactions. But IOTA led attempts to secure IoT is far from commercialization. This is posing two folds challenges in growth of IoT. First adoption is slowed. And second because adoption is slow, investment to overcome security challenges is so far scanty.

4. Lack of standardization : Engineering can attain quality only if two essential conditions are met. These are - existence of standards in design and readily available parts and parcels to build according to design. In the context of IoT this essentially means, every IoT device, its firmware, security are built to standards and have been matured in the field. But IoT systems of today lack standardization and maturity as technology is fast evolving. Inherently in every branch of Engineering, design & architecture can be challenging but it works because we designs from standards and readily available and proven technology. Evolution of IoT Technologies in last six to seven years didn't converge to any standard and neither deployment has reached such level of maturity. Therefore, one needs very good system engineers to analyze business cases and cover every vulnerable corners. Unfortunately system engineers in the area of IoT are not available in plenty because of lack of large scale deployment.

5. Continuous deployment strategy : What is ignored in many IoT technology and business design- how human beings will consume and use valuable IoT data or actionable intelligence. This poses two fold challenges. First one I already discussed- how and why should an organization would accommodate the changes necessary to follow actionable intelligence. And secondly this means one needs sizable amount of beta deployment to receive feedback from the customers before product is matured in terms of user experience and adoption. Hardware design and infrastructure part can freeze very quickly. But analytics needs continuous deployment and improvement. Had that been 100% in cloud, it should not have been challenging. But in the days of edge analytics for both middleware and firmware need continuous update. This can end up in nightmare unless thought thoroughly in advance.

1-5 will help you to understand the challenges faced by Internal IoT team of large organization.

Firstly, most of them have very well defined product deployment, product line management (PLM) procedure for releasing a product. Those ISO driven approach will not work in the area of IoT because a product manager can define only 60-90% of the product at beta release. Rest 40-10% can be understood only after beta customers provide adequate feedback. A large scale beta is extremely important for successful IoT. But it's extremely difficult for large corporations to adopt a flexible beta strategy like Google because of uncertainty and draining of resource and budget.

Second mistake by most of the Internal IoT team- hiring of different skills that can't talk to each other. They will hire data-scientists, architects, middleware, cloud, Big Data experts etc. They will design in Silos and then they try to stitch each. That won't work well in IoT. Because every one depends on others and they can't design without knowing every layer. Now that is difficult. How can you expect a data scientist to know vertical engineering knowledge as well as deepest secret of system engineering? We are faced by these daunting challenges in IoT because there exists no standard to make their life easy.

Third mistake-assumption that IoT holds prospect for too much of revenue and so quickly! Look at GE. They burned two billions on Predix too quickly. Such big market in IoT can't be created overnight. Not in two years. It takes more than two years just to do a closed beta successfully in intelligent IoT. Companies have to take long term evolution strategy for IoT. It takes a Five Year planning and not two year sprint to get any IoT product done properly. When I say properly meaning- something working in scale and with full satisfaction of the customers.

Consequence of all the above have been catastrophic on companies who have burned too much of money into IoT too quickly. GE and Verizon are good example. I also know many other IoT initiatives from large companies which met similar fate. Either they have shut down their IoT division or scaled back to 5% level. Human cost has been staggeringly high. Many of the young engineers who jumped into the IoT bandwagons with high hope for a great career in a great field had to go through mass scale lay-off and shutting down of their division.

Therefore before starting any IoT project both the Engineers and Product Line Management have to be realistic about potential revenue and scale of the issues associated with it. They need to hire experienced team or leaders who have deployed IoT as large scale and lived through its pain and automation. Once they have formulated their system architecture for IoT, it is always better to invite experts and review the plan and supporting manpower.

One has to understand difference between Industrial 4.0 with 3.0 to set the expectation right. In 3.0, network and devices are robust. Industrial Networks do not allow CSMA/CD kind of protocols as that will create contention and packet loss. Where as for most part, Industrial 4.0 has to run on cost effective electronics on public internet backbone. Also, for 3.0, objective and ROI is simple to calculate. It is replacement of operational manpower. For 4.0, ROI can be highly complex and complicated.

Therefore, one has to understand unlike other areas of Engineering, IoT is not yet a commodity. It's the most complex web of Engineering which has not matured yet.