Let me walk you through those four domains in a bit more details. The way I would like to do this is the following. For each domain, I will share with you some popular myths, the ones I have heard from clients, friends, or the press. And then, I'll tell you my view on each of them. All right. Let's start, the first myth is deeply anchored in many companies, big data as a technology question, and that's not true. In the last few years, I have gathered numerous horror stories about lost investments in Hadoop platforms or data lakes. Hundreds of millions, actually sometimes north of one billion dollars invested without a business objective in mind. As a result, all thrown out and restart from scratch. So, if you see that a company's data and analytics vision is about getting the latest and greatest technology in town, you know the risks. The vision doesn't need to be as ambitious as Google's organizing the world's data. But it needs to resonate with your business objective, your business outcome. Let's now move to data usage myths. The first one is what I would call the myth of gut feel. This is one of the cultural hurdles to using data as a decision making tool. Let me be clear, I don't deny that sometimes business leaders can have strong intuition or be visionary figures in their markets. There's even some interesting neuroscience research behind this, but you can't run all departments and take all decisions based on everyone's intuition. And you certainly shouldn't do it when it's a critical decision. Let me give you another horror story. Ron Johnson, former senior vice president of retail operations at Apple, he became CEO of J.C. Penney in 2011. He very quickly implemented a no discount strategy based on his gut feel. His colleague suggested to test it in few stores before to lend it out but it didn't happen. It turned out his gut feel was fundamentally wrong, and J.C. Penney shares have declined 51 percent. To create the right use cases for data usage, it is necessary to promote a data driven culture in the first place. The second usage myth is one we have discussed with John Rose in module three, it's about trust. And it's a myth you can hear in large companies, in startups, and in the media. Data privacy is a problem for the old generation. Well, privacy is actually a key issue for all generations. And it's not because a teenager is more comfortable posting photos on Instagram, that he doesn't care how those photos are being used. And when we look at multiple surveys across different countries, we see that there is less variation between generations than types of data. People are more sensitive to privacy of their healthcare records than they are to their buying habits. Overall, companies who want to be able to collect and use their customers data, should protect against misuse, regardless of which generation or which group of people that they are targeting. Let's now talk about the data engine, about technology solutions, data gathering, skills required, and the link to business processes. Data and analytics technology has seen the rise of many buzzwords in the last few years. Hadoop platforms or in the memory analytics became so trendy, that some managers started thinking that they will replace all existing infrastructure. Well, planes didn't replace cars because they both answered different trade-offs. Hadoop or data lakes will not replace data warehouses, because the latter are still effective for structured data where parallel processing is not needed. And in general, we shouldn't be looking at technology solutions from the novelty alliance, newness, and lack of maturity if anything, come with certain risks. What is important is the trade-off between the functionalities you need, and the cost you're willing to pay. On the data side, I often hear statements like I have a B2B business or I am in support function, I'm not as lucky as those businesses with tons of end customer data. This is clearly a myth. Big data has numerous applications, valuable applications, in Operation analytics, like procurement optimization, predictive maintenance, logistics optimizations or others. A couple of years ago, we did a warehouse network optimization for a procurement department in Southeast Asia. It led to 20 percent savings on their logistics operations with no additional CapEx investments. The third myth in data engine is about people. I like to call it the perpetual search of data scientist unicorn's. That one person that will combine deep IT know how, superior analytics, skills, great business expertise, and a strong ability to influence decision making. I can tell you those people are very rare. So, don't build all your hopes on them. Setting up a strong team including all different skills in the way is the way to make up for missing unicorns. Especially if close collaboration within the team is ensured through an agile way of working. You have heard about that from Sonia in the previous video. There's one last myth on data engine that I want to share with you. It goes something like this, once you see the value of data and analytics, change will happen. The reality is although proof of value is a critical component, it is not sufficient to make change happen. I remember when we helped a telecommunications company build a data driven channel optimization capability, the proof of value took a couple of months. Sales growth actually increased by 20 percentage points. But to make this impact sustainable, it took a few more months to create the new central team to ensure that it has the right interfaces with stores operation teams, and reseller management teams, and to integrate the new approach to the company's decision making process. And this was the key, data analytics should not live in silos, they should be embedded in the daily business operations. There are two more myths I want to share with you on the fourth domain, the data ecosystem. The first is about this tendency to only trust what is homemade. The belief that a company should do everything by itself. The truth is, it's much more efficient to rely on a wider data ecosystem for both data exchanges and analytic support. There's a great story here from NASA, the space agency. They wanted to develop an accurate prediction model for Solar Particle Events, SPEs. And they have tried for more than 40 years with limited success. In 2010, they had the brilliant idea to crowd-source a better predictive model. They found a solution within three months, and that solution was 50 percent better than their existing model at a cost of $30,000. NASA is not the only one, Netflix did the same thing a few years ago to improve their recommender system. Now, the last myth I want to discuss says that it is hard to openly share data beyond your organization boundaries. Okay. Well, this is not fully a myth, not yet at least. It is hard to build the right relationship framework that allows seamless sharing of data. We know it creates disproportionately more value when you can correlate different sources of data to find new and useful patterns. This has encouraged many governments to adopt this Open Data culture. But if the data is part of your business competitive advantage when you are building an AI capability for example, then it is very hard to defend full openness. Companies willing to mutually share data, are usually get it around this hurdle by creating joint ventures as an example. There are also some ongoing initiatives in Singapore, for example, to build data marketplaces, platforms to facilitate discovery and exchange of data sources between industry players. What should you takeaway from all those real life stories? A company needs to start its data and analytics journey with a clear vision closely linked to its business priorities. The vision needs to translate into tangible use cases, that can serve as a compelling proof of value. Big data does not need magical capabilities, teams can be created with the right combination of skills to replace a data scientist unicorn for example. Businesses shouldn't feel obliged to do everything in-house, they should partner with their ecosystem to enrich both data sources and analytics capabilities.
