Digital roadmaps are conceptualized very differently across organizations. Some go for “digitisation, digitalisation, transformation,” some go for “automation, augmentation, amplification.”, and others go for “products, platforms, ecosystems.” In my opinion, strategy has little contribution to an organization’s eventual digital success. Success should be defined by our market performance and not how intellectually stimulating our PowerPoints are. In this digital age where consumer data is so easy to collect, we can measure our success almost immediately. Thus, we should not spend time trying to argue how solid a piece of strategy is but organize our teams to quickly get to the point of experimenting with actual customers. No one is better than your customers in leading you to success.

Another big hurdle that I see is legacy systems and isolated IoT sensors/devices. Many organizations embark on data quality programs, re-working core systems, and unifying their data architecture, which costs a lot of money, time, and resources. The irony is that once they are done (usually never done to perfection), these technologies are obsolete again… This is why today, we are strongly recommending a “lego-block” approach where anyone can bring their own data, bring their own devices, bring their own systems, and we use a Data Fabric as a real-time data translator and an Orchestrator to compose independent modules into new business capabilities. This architecture is what Gartner terms “Composable Intelligent Businesses,” and this is something worthwhile to explore further.

The crux of the issue is not emerging technology or any technology for that matter. An organization that has a culture and organizational structure that supports innovation often leads the pack in the successful implementation of emerging technology. This core capability is organizational in nature: The freedom to experiment rapidly.

Organizations should never chase after technology. In today’s digital era, companies thrive by being the fastest to deliver what consumers want rather than what they (the companies) can do. Given that consumer preferences and the competitive landscape are changing so rapidly, the only way for companies to catch up is to experiment with multiple use cases as rapidly and cheaply as possible by pruning lousy features, improving promising features, and re-orchestrating good features in an iterative manner.

It is through the luxury of rapid experimentation that teams can test out all kinds of solutions for their customers: from just process innovation to mainstream tech to emerging tech, and the solution that they eventually choose must be the one that delivers the strongest customer value proposition at the lowest cost to the company. From my experience, emerging tech is often not the best solution, but unless the team has the luxury to try them all out, they would never know.

Note: Customers can include consumers, as well as external and internal stakeholders. As long as they are the beneficiaries of our products or services, we see them as customers.

The “luxury” of rapid experimentation is not a function of company wealth. It has to do with how the company works - namely, who makes product and service decisions (it should be the people closest to the market), and how teams are incentivized to create the best solution and not “just the most convenient solution that works.” Give our teams space and empowerment to experiment, and naturally, emerging tech will be included in one or more prototypes.  

Thanks for reaching out to me through USAEI. Your experience with analytics is not uncommon. Most organizations have recently complained that they don't see value from their analytics, machine learning (ML), deep learning (DL), and even artificial intelligence (AI) endeavors.   

The problem does not lie with the analytics technology. Rather, analytics is only a tool that guides, validates, automates, augment, and amplify business decisions. The problem lies with how most organizations define the business value that they want analytics to solve. 

These are some of the common problems that I have seen, and one leads to another:

1. Business managers do not have a concrete idea of what they want to solve and hopes that analytics can help them ask the right questions. Analytics, ML, DL, AI are tools, they are incapable of asking questions. However, given the right questions, they can help you find answers and insights that humans cannot match. Just as we don't expect excel to generate business hypothesis, we can't rely on analytics, ML, DL, and AI to do the same. They are just a very powerful tool, not a domain expert.

2. Business managers have a poor definition of value and impact. A good value proposition is a concrete statement with 3 components as a fill-in-the-blanks statement: As a <which type of customer>, I want <what features/products/services>, to achieve <what objectives do I really care about>. Type of customers can be other businesses, consumers, internal staff, management, and any other stakeholders. If we find that we cannot fill in the blanks with a convincing statement, then it means that the idea is either too vague, broad, naive, confusing, or not valuable. Very often than not, once we have a good set of value propositions, we realized that we don't need analytics, ML, DL, and AI to solve it. Most business value can immediately be realized with a better process, incentive design, platform business model that directly address what the customer wants and not what the business can do.

3.  Using analytics, ML, DL, and AI like how we use excel - reporting. The value of analytics, ML, DL, and AI does not lie in cost-cutting type of automation or reporting. Its augmentation and amplification, meaning that they are designed to help us deliver value by making better decisions given the business questions. The corresponding output of analytics, ML, DL, and AI is actionable insights. However, for them to derive actionable insights requires (1) and (2) to be defined first, else its like trying to discover if there's something that might interest somebody, somewhere in the data.

4. Next, the business competency of the data analyst. There is a hierarchy of competency for data analyst and it starts with experience and communication. A good data analyst is also one with sufficient business intuition to understand what the users of the solution needs and how we can find them in the data using the latest tools. If a data analyst is always behind the screen and does not spend a significant amount of time talking to stakeholders and conducting small demonstrations to solicit feedback, then he will never be able to create impactful actionable insights.

5.  Lastly, the technical competency of the data analyst. Technical and domain experience are complementary. Strong technical experience will make up for domain experience and as their domain experience increases through experimentation, their value doubles. Knowing how to use analytics software is not a technical expertise. Technical skillsets revolve around knowing how to make data speak, how different models work, how to create visualizations that communicate multiple ideas almost instantly, and how results from one experiment lead to another, and another, before finally uncovering actionable insights. Thus, in truth, there are no paid analytics software in the market that can outperform the best suite of open-source tools in the market if the data analyst is competent. Using off-the-shelf analytics software simply allows data analysts to hide their insufficiencies behind automated features.

As you would have realized by now, any analytics projects that we task, regardless whether other organizations brag about it or being written by big three consulting agencies, they are ultimately irrelevant until we have points (1) to (4) sorted out.

Jack

Having a lot of data is not the objective. What we really want is useful data. For data to be useful, it must be both accurate and representative. Data integrity issues such as duplicated entries, multiple identifiers for the same item, incomplete fields, and selective recording may cause an analytics system to recommend the wrong decisions.

SMEs must recognise that the new drivers of digital transformation are strategic in nature, with technology playing a secondary role as enablers. They need to first figure out how to adapt their business models with the new drivers, before evaluating what technology will best apply.

Digital Transformation is a set of new business drivers that obliterates barriers to industries that were previously built on old drivers of success. These drivers include consumer networks and advocacy; “co-opetition” and the platform business model; data, analytics, and artificial intelligence; innovation, rapid experimentation and agile practices; value proposition-based decision making and digital-savvy leadership.