Digital Transformation: the Elephant in the Python
Lately I've been thinking a lot about the latest fad in business thinking - digital transformation. As technologies become more pervasive, and our ability to gather and process a lot of information increases, it makes sense to think about how digital solutions may change the way we work, either by replacing monotonous tasks, automating entire business processes, anticipating future trends and hundreds of other ways.
As more companies place more sensors in their products, the products will start reporting on their mean time to failure and maintenance needs. Companies will gather data from those sensors and provide feedback on optimal usage, maintenance and repair or replacement needs. The Internet of Things (IoT) will mean that many, many devices are connected to the internet and providing a constant stream of data that can be analyzed, so that we can learn more about any factor of our supply chain, distribution chain and consumer consumption of products and services. When you add in the promise of blockchain, it's possible that all of those transactions will be conducted in a fully automated and transparent ledger. What's not to like?
Taking the curmudgeonly approach
This digital future sounds incredible, because it is. We can now automate far more than ever before. We have more data at our fingertips than ever before. With that data we should be able to schedule people and machines more effectively, anticipate future shifts and reduce costs and cycle times. We should know more about our entire value chain than ever before. The real question is: can all of this technology and data provide the insights it promises? Is there a weak link in all of this promise? I think there is.
Digital transformation is already happening. It's not something that will arrive all at once, but piecemeal throughout your business. You've already installed some IoT devices, and probably have a relatively robust ERP system to govern your internal processes. To some extent you are already on your way to becoming a 'digital' business. But there is more you can do, with additional sensors that capture data about customers, or blockchain applications that track the provenance of products and services. You can and will install AI or machine learning algorithms to help you improve the efficiency of your operations. As I said, this stuff is already happening, and it is a double edged sword.
The promising edge of that sword is integration and the availability of data and, with the right processing, new information and insights. The difficult edge of that sword is integration of all the data producing equipment and the ability to normalize the data so that it is all integrated and useful. And this is no small challenge. Once every IoT device, smart device and distributed system comes on line, they will generate terabytes of information, and often in different data configurations. Since there is no standard, someone must normalize all of this data and decide which data is truly valuable, and how it will influence or impact AI and ML algorithms. You could literally be awash in data from thousands of different devices, all of which are producing data that has potential value, but in different data standards and configurations.
Who does the integration?
Think about your current integration challenges. Today many corporations have manual consolidations across different IT systems because it's easier for humans to review and consolidate financials or other data sets, rather than create rules about how the data should be aggregated. If you have challenges with your existing operational data, collecting it, aggregating it and normalizing it, what do you think will happen when hundreds of new data sources - both internal and external - come on line, and new systems like AI or ML try to start making sense of this data?
Worse, an already overworked and overwhelmed team - the IT team - which in most cases is simply trying to keep the older operational systems afloat - will be called on to normalize and integrate all of this data, and ensure that the AI and ML systems get good, clean data to make assessments and projections. Oh, and they'll need to install all the hardware and software, determine data flows and continue to run the regular IT systems that keep the company running. Why is no one talking about the big challenge of data integration that is required for digital transformation?
The elephant in the python
Data integration isn't just the elephant in the room that no one is talking about, it is the elephant in the python. Digital transformation assumes that the data that is produced at any place in the business is instantly available to everyone, creates maximum efficiency and helps people and machines make better decisions. For that to happen data needs to flow accurately and freely, and be free of conflicts or errors. All systems need immediate access to the data and AI and ML need to be able to process the data and make decisions in real time. What I just described is a fantasy for most large corporations today, and it will be a fantasy for quite some time. Data simply doesn't flow that way, and today we are talking about a mere trickle of data as compared to rivers, oceans of data that will spring forth as the IoT devices begin to pop up everywhere.
Most firms don't have the ability to manage the volumes of data we are talking about, don't have the ability to normalize the data and certainly don't want to introduce all of these diverse streams of data into any decision making process or algorithm without a lot of oversight.
Consultant Job Creation
When people tell you digital transformation will make your company more efficient, be wary. It strikes me that it will create massive data traffic jams internally, leading to the need for far more consulting labor to straighten out the data flows. It will require far more people to sit in judgement of the data, to ensure the data is valuable and useful, to keep algorithms from using incorrect or flawed data sets to make decisions. Digital transformation is the pot of gold at the end of the rainbow for many hardware and software producers, who are overstating the benefits (of which there are many legitimate benefits) and vastly understating the challenges having to do with integration of data, the normalization of the data and the sheer volume of the data.
Does IoT, blockchain, AI, ML, sensors and other technologies have a place in your business? The answer is unreservedly yes. Should you be concerned about the use of these technologies and the impact on your business? Again, yes. Where is the biggest challenge? I'll stipulate that most companies simply don't have the ability to manage the amount of data that will be created, and can't make that data (and the more important information and insight that should be gleaned from the data) available to the right people to make the right decisions quickly. What's the value of digital transformation if it does not make the business more productive?
As more companies place more sensors in their products, the products will start reporting on their mean time to failure and maintenance needs. Companies will gather data from those sensors and provide feedback on optimal usage, maintenance and repair or replacement needs. The Internet of Things (IoT) will mean that many, many devices are connected to the internet and providing a constant stream of data that can be analyzed, so that we can learn more about any factor of our supply chain, distribution chain and consumer consumption of products and services. When you add in the promise of blockchain, it's possible that all of those transactions will be conducted in a fully automated and transparent ledger. What's not to like?
Taking the curmudgeonly approach
This digital future sounds incredible, because it is. We can now automate far more than ever before. We have more data at our fingertips than ever before. With that data we should be able to schedule people and machines more effectively, anticipate future shifts and reduce costs and cycle times. We should know more about our entire value chain than ever before. The real question is: can all of this technology and data provide the insights it promises? Is there a weak link in all of this promise? I think there is.
Digital transformation is already happening. It's not something that will arrive all at once, but piecemeal throughout your business. You've already installed some IoT devices, and probably have a relatively robust ERP system to govern your internal processes. To some extent you are already on your way to becoming a 'digital' business. But there is more you can do, with additional sensors that capture data about customers, or blockchain applications that track the provenance of products and services. You can and will install AI or machine learning algorithms to help you improve the efficiency of your operations. As I said, this stuff is already happening, and it is a double edged sword.
The promising edge of that sword is integration and the availability of data and, with the right processing, new information and insights. The difficult edge of that sword is integration of all the data producing equipment and the ability to normalize the data so that it is all integrated and useful. And this is no small challenge. Once every IoT device, smart device and distributed system comes on line, they will generate terabytes of information, and often in different data configurations. Since there is no standard, someone must normalize all of this data and decide which data is truly valuable, and how it will influence or impact AI and ML algorithms. You could literally be awash in data from thousands of different devices, all of which are producing data that has potential value, but in different data standards and configurations.
Who does the integration?
Think about your current integration challenges. Today many corporations have manual consolidations across different IT systems because it's easier for humans to review and consolidate financials or other data sets, rather than create rules about how the data should be aggregated. If you have challenges with your existing operational data, collecting it, aggregating it and normalizing it, what do you think will happen when hundreds of new data sources - both internal and external - come on line, and new systems like AI or ML try to start making sense of this data?
Worse, an already overworked and overwhelmed team - the IT team - which in most cases is simply trying to keep the older operational systems afloat - will be called on to normalize and integrate all of this data, and ensure that the AI and ML systems get good, clean data to make assessments and projections. Oh, and they'll need to install all the hardware and software, determine data flows and continue to run the regular IT systems that keep the company running. Why is no one talking about the big challenge of data integration that is required for digital transformation?
The elephant in the python
Data integration isn't just the elephant in the room that no one is talking about, it is the elephant in the python. Digital transformation assumes that the data that is produced at any place in the business is instantly available to everyone, creates maximum efficiency and helps people and machines make better decisions. For that to happen data needs to flow accurately and freely, and be free of conflicts or errors. All systems need immediate access to the data and AI and ML need to be able to process the data and make decisions in real time. What I just described is a fantasy for most large corporations today, and it will be a fantasy for quite some time. Data simply doesn't flow that way, and today we are talking about a mere trickle of data as compared to rivers, oceans of data that will spring forth as the IoT devices begin to pop up everywhere.
Most firms don't have the ability to manage the volumes of data we are talking about, don't have the ability to normalize the data and certainly don't want to introduce all of these diverse streams of data into any decision making process or algorithm without a lot of oversight.
Consultant Job Creation
When people tell you digital transformation will make your company more efficient, be wary. It strikes me that it will create massive data traffic jams internally, leading to the need for far more consulting labor to straighten out the data flows. It will require far more people to sit in judgement of the data, to ensure the data is valuable and useful, to keep algorithms from using incorrect or flawed data sets to make decisions. Digital transformation is the pot of gold at the end of the rainbow for many hardware and software producers, who are overstating the benefits (of which there are many legitimate benefits) and vastly understating the challenges having to do with integration of data, the normalization of the data and the sheer volume of the data.
Does IoT, blockchain, AI, ML, sensors and other technologies have a place in your business? The answer is unreservedly yes. Should you be concerned about the use of these technologies and the impact on your business? Again, yes. Where is the biggest challenge? I'll stipulate that most companies simply don't have the ability to manage the amount of data that will be created, and can't make that data (and the more important information and insight that should be gleaned from the data) available to the right people to make the right decisions quickly. What's the value of digital transformation if it does not make the business more productive?
Voted 2nd best innovation blogger two years in a row!
0 Comments:
Post a Comment
<< Home