Updated: Apr 23, 2021
Digital transformations have become a global trend in recent years. To be clear, in mainstream understanding, the term means to increase the use of data, which can then help us to build “smarter” machines, predict the future, dig out insights, eliminate human errors and maximize efficiency. However, according to the stats released by Boston Consulting Group (BCG) and McKinsey & Company, only about 30% of digital transformation projects ended up successfully. The result keeps us wondering: What are the key issues to account for such high failure rate? And more importantly, how can we resolve these issues?
In this article, 5 major problems that many companies encountered during digital transformations are introduced. We also provide useful suggestions on how to overcome them.
Problem 01: Having no reasonable data strategy.
Collecting insights from data (either using AI or human data scientists) is the essence of all digital transformations nowadays; it allows ones to have deeper understanding about their own products, equipment, customers and the market, which then leads to better decisions or more accurate predictions. In order words, the digital transformation today is basically a process to embrace data-driven way of thinking.
In order to enjoy the fruits of data, however, companies must adopt a reasonable data strategy first. Generally, when speaking of data, the following issues should come into your mind immediately:
What data do you want to collect?
All digital transformation must have a purpose (that is, the problem you’d like to solve; we’ll cover more on this topic in the next section), and the purpose will determine what data you should collect since different data show different insights.
For instance, to understand consumer behaviors on an e-commerce website, we must know information such as “how long have people spent on a certain page in average” and “in which step in the payment process do most people give up”. And if a farmer wants to increase production using digital methods, he or she will need to monitor data such as sunshine, humidity, etc.
How to collect these data?
Following the point said, different data will have to be acquired in different ways. To access e-commerce data, for example, we can utilize tools such as Google Analytics, Matomo (formerly Piwik), Kissmetrics, and so forth. As for the data required in a farm, we must resort to sensors designed specifically for collecting a particular type of input (e.g., light, sound waves, and moisture).
How good are these data?
As the saying goes, “garbage in, garbage out”. Data will provide no value (or even harmful influence) when they are unclean. If a decision maker makes a decision based on low-quality data, he or she will likely receive negative outcomes. Such event can lower the decision maker’s faith in data-driven approaches, hence hampering digital transformation.
To keep this from happening, one must perform preprocessing before exploiting any data, and that includes techniques such as missing / duplicate data handling, one-hot encoding, feature scaling, etc. We’ll enlarge on these topics in our future articles.
When sensors are involved in the data collecting process, it’s also crucial to ensure that you’ve equipped and calibrated them correctly, and do check if they are functioning normally on a regular basis. As for those who use website traffic monitoring tools, be sure to filter out the IP addresses from your and your colleagues’ devices so that the traffic data collected are representative.
Who can access these data?
After getting valuable data, it’s important to store them in a secured location and ensure that only the authorized personnel can use them in a reasonable way. Such issue has become significant in recent years due to the following reasons. First, more and more people are working remotely, which means that workers may be accessing corporate data with their personal computers instead of company’s ones. Second, due to the trend of IoT, more and more devices are linked together, and that inevitably increases the risk of unauthorized connections. Accordingly, how to spot insiders and how to consolidate IoT security have become hot topics that today cybersecurity experts focus on.
Also worth mentioning, it will take a lot of effort for human beings to overcome our own instinct, which is to trust past experiences over data. Often, the attitude of the key decision maker(s) in a company plays the most crucial role. Those leaders who allow their employees to question their decisions using data usually have a higher successful rate in digital transformation than those who do not.
Problem 02: Your digital transformation project is purposeless.
As most revolutions, digital transformations involve breaking the old paradigm and establishing a new one. The process is like finding the new world while sailing on the ocean; one can expect how difficult it will be if we have no clue about which direction to take. So, before you decide to transform, you must find a goal to attain to first; otherwise, you will likely confuse your employees for nothing by messing up how they used to do things. In digital transformation, the goal is usually about solving a problem that can be resolved through one or more data-driven solution(s) more effectively; not until such problem is determined can one plans out what to do next accordingly. Note that the more specific your goal is, the more likely your business will benefit from digital transformation.
Let’s see a practical example. Say, a factory owner would like to “reduce the downtime loss caused by equipment malfunction”. Since this is a very specific goal, we can then work on how to achieve it. One potential solution is to (1) deploy an AI algorithm capable of detecting the subtle signs of malfunction (such as unusual noises) that human senses cannot capture, and (2) connect all machines of interest to a monitoring app using IoT techniques so that one can get instant alert before these machines actually break down.
With a specific goal in mind, one can not only have a clearer idea about what to do but also better evaluate the success of a digital transformation project. Takes the case said as an example. We can easily tell whether our new approach is beneficial or not by tracking the downtime rate and the money spent on maintenance (since our AI algorithm can foresee failures, we can have a better control over when, and maybe how, to fix our machines).
Remember, transformation is not always good since it induces drastic instability. To harvest its merits, you must utilize it at the right moments, that is, when you encounter a problem that can hardly be resolved through the old ways.
Problem 03: IT and OT people do not work together.
The gap between IT and OT people is a common and intractable issue, especially for factories. In digital transformations, the impact of such gap is also significant considering that IT people are the ones who cook up solutions while OT people are the ones who use them. Thus, without proper communications between the two parties, IT people can hardly come up with something applicable in an actual environment, which then leads to the failure of the whole project.
To overcome such issue, the support from the business owner plays a very important, if not the most important, role. He or she should not only have the ability to bridge the two parties, but also give actual power to the managers who are responsible for the transformation. However, a lot of business owner fails such functions since he or she does not have relevant knowledge, and that brings us to our next problem.
Problem 04: Lacking relevant knowledge.
Digital transformations involve concepts and techniques from a very wide range of areas, including data science, cloud / edge computing, artificial intelligence (AI), internet of things (IoT), telecommunication (e.g., 5G, or even 6G), etc., and therefore it’s very difficult for a business owner with no technical background to grasp all of them. If that’s the case, business owner will hesitate or fail to make correct decisions regarding digital transformation because they do not have sufficient information to work with.
In order to avoid this problem, business owners nowadays should realize the important to include people with different skills, talents and backgrounds in their team. Only through that can one have correct messages about everything involved in a digital transformation project. It’s also helpful to employ a capable CTO (Chief Technology Officer) so that one can deal with issues related to technologies more smoothly.
Besides, encouraging employees to take relevant courses or resorting to consulting services are also good ideas.
P.S., To proffer you a rough picture, we include a graph and some simple explanation of all key technologies regarding today digital transformation in the appendix. We sincerely hope that it helps.
Problem 05: Employees do not cooperate.
Although technologies are the critical players in digital transformation projects, the communication skills of the employer are the key factor to decide whether such projects will be successful or not. When facing great changes, it’s only natural that employees will have concerns in mind (this is especially true with tech such as AI – people are afraid that they will lose their jobs due to automation). In such scenarios, the words from the business owner will become very crucial.
The following is one bad example of employee-employer communication that actually took place in a factory. In order to be impressive, a general manager led his old customers to one of the production lines, where his employees were currently working, and said the following words out loud: “by next years, this production line will become fully automatic!” You can imagine how such declaration can trigger turmoil inside the factory.
So, what can a business owner do to convince his or her employers to participate in digital transformation? One thing to do is to emphasize how learning new skills (e.g., data mining, building neural networks, etc.) can be beneficial to them, or how new technologies are making their jobs easier (not replacing them). It’s also helpful to set up mechanism to evaluate employee’s performance based on what they’ve achieved in the digital transformation projects and offer rewards accordingly. Although some may not like it, this is actually a highly effective approach to motivate people.
Conclusions: Digital transformation is a revolution of minds!
The saying, “data is the new oil” accurately describes the reality we’re facing. With more and more people embrace the digital lifestyle, the information that data can tell us is literally unlimited. To harvest the benefit, however, one must first adopt the data-driven way of thinking, and that can be very different from how we consider things in the old days.
Technically speaking, both human brains and computers store and take advantage of data, but they do it differently. To be specific, although cerebrums can learn from a small amount of data, they often give an unreasonably high weighting to an unrepresentative sample (think about how powerful first impression is to our judgment on an individual). Machines, on the other hand, tend to treat each sample fairer and more equally. As a result, even though they may require more data to form a conclusion, such conclusion is often more reliable than human experiences (this is on the premise that the data machines take in are in high quality). Therefore, in order to make a digital transformation project successful, the first thing that a business leader or key decision maker should do is to stop making decision based on experiences and start listening what data has to say.
But it won’t be enough if there is only one person adopting the new philosophy. So, the next thing an employer should do is to pass the spirit down to all of his or her employees. That’s where human factors shine! As one may know, many projects fail due to the people involved show poor communication skills, and that should remain to be true even with the advance of technology (in fact, communication skills should become more and more important in our opinion since the world is becoming more complex, which render one individual hard to know and do everything but most resort to group work to achieve a goal). Thus, learning how to communicate with other people effectively should be an imperative mission for all of us.
All in all, essentially, digital transformation is a revolution of minds rather than just technology. It’s certainly not easy, as most transformations, but it can significantly strengthen what we can accomplish if done correctly. Hopefully, this article can help you to spot some problems in your own projects and have some ideas about how to resolve them. Good luck!
[Appendix] Key technologies of digital transformations in one graph
On the basis of digital transformation is 5G (or even 6G) technology, which provides features known as enhanced mobile broadband (eMBB) and ultra-reliable and low latency communications (URLLC) that enable large data transmission in an extremely high speed (from 50 Mbps to over 1Gbps) as well as ignorable latency (≤ 1ms). It can also realize massive machine type communications (mMTC), which can connect more than 1 million devices and therefore plays a very crucial role in IoT. To learn more about 5G, please refer to another article of ours.
In the future, all devices (including personal computers, laptops, mobile devices, self-driving vehicles, and other types of machines / sensors) should link together to form a comprehensive internet of things (IoT). This not only allows us to monitor and control all of our devices remotely, but also permitting data to be shared more quickly and effortlessly among different types of devices.
It’s important to know that all the devices aforementioned are continuously producing new data, which can be very valuable if processed right. Since the number of these machines is enormous, together they give us datasets so huge and complex that cannot be handled with traditional software or approaches, and that’s why we need novel methods capable of handling these big data.
To better store, manage and analyze big data, we may want to transmit them to a cloud platform that is bolstered by specialized servers or powerful data center rather than keeping them local. There are three types of cloud in general. Public cloud is platforms built by third-party companies and available to the public through the Internet; three of the most well-known public cloud services are Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP). In contrast, private cloud is established by the company in need, and therefore it can only be accessed and will have to be maintained by the very company. As for hybrid cloud, it simply means that a company uses both public and private clouds. Note that although building a private cloud can be expensive, it is necessary in certain scenarios, such as those when patients’ personal information is involved.
Since cloud computing (the processing and calculations performed on cloud) is significantly more powerful than local computing, we usually train our artificial intelligence (AI) models with machine learning methods there because these models are usually data-hungry. However, if we leave our AI models on cloud, its response time may be too slow for practical use, especially for the situations where no-latency reactions are required (e.g., autonomous vehicles), since our data center may be far away from where we actually are. Hence, we must deploy trained models on edge devices that are located just around the workspace or factory; these devices can help us performing simpler computing (i.e., edge computing) and at the same time satisfying our request of low latency.
Footnote: Neurozo Innovation provides viewpoints, knowledge and strategies to help you succeed in your quest. If you have any question for us, please feel free to leave a comment below or e-mail us. For more articles like this, please follow our Twitter, Facebook page, or LinkedIn. Thank you very much for your time, and we wish you a wonderful day!