Bigger data, Better decisions – How to be smartly data-driven?
Big data was supposed to be the “new oil” and the smartest thing businesses can have. But researchers claim data is not of much help if you don’t know how to use it. Journalist Ville Blåfield examines what type of cultural shift it would take to be smartly data-driven. Illustrations Jarkko Hyppönen.
“We call this the problem of big data”, NASA scientists wrote in their publication back in 1997.
Space technology experts had come across something they described as
”an interesting challenge for computer systems: When data sets do not fit in main memory, or when they do not fit even on local disk, the most common solution is to acquire more resources”.
As far as known, this was the first instance a phenomenon was called “big data”. The term has spread like wildfire since then – at more or less the same force as the increase of data on servers and networks around the world.
In 1999, researchers at Berkley estimated that data produced around the world amounted to 1.5 billion gigabytes. The current estimate is 2.5 quintillion bytes of data in the course of a single day – a figure you arrive at by adding eighteen (18!) zeros after 2.5.
Modern, digitalized life creates vast data masses, and obviously “big data” is seen as a problem, fascinating challenge, and an incredible opportunity.
Big data has been proclaimed as the “next oil” (Fortune magazine, 2012) as well as the “next coal” (Guardian newspaper, 2016.)
Big data offers answers that were previously down to guesswork in areas like forecasting human behavior, monitoring trade, and business assessment and planning. In this respect, understanding and using big data has become self-evident in every field: of course people wanting to develop their business should use it!
But in one respect we are none the wiser than the NASA scientists back in the 1990s. Managing big data continues to be a challenge, and making use of the data masses wisely – with genuine understanding – is even harder.
There are epic success stories from around the world of businesses harnessing customer or user data to steer their strategy in a smarter way. But another recurrent story is that of companies repeating the importance of big data in the boardroom without anyone in the conversation really grasping what it actually means for the company.
As the joke went at a seminar for communications experts in the fall: “Big data is like sex in junior high: everyone talks about it, but few have experimented”.
Henri Schildt is a tenured professor with a joint appointment at the Aalto University School of Business and the School of Science, Department of Industrial Engineering and Management. He is currently heading two research programs on data-driven management and the management of digitalization.
With PhD and M.Sc. from Aalto University School of Science, Schildt has followed the development of information systems and the use of big data in management for more than a decade.
“When big companies like IBM began to actively advertise and sell their big data analytics tools a decade ago, little research existed on the use of big data in management.”
In principle, applying data analytics is nothing new for those operating on the interface of business and science. Science and research have always organized data and aimed to understand it.
“Statistical methods are tools for understanding the world. This began to shift from researchers to companies. Data analytics methods just transferred across extremely slowly and selectively.”
According to Schildt, an “operational” use of big data has taken root in businesses over the last decade. Similar analyzes with smaller data sets have been mainstay of operations research for decades. Consumer purchases help forecast future sales, product development is fine-tuned according to user feedback, and areas for savings are detected. The next level of understanding would be more abstract, or tactical, as Schildt says.
“Companies usually still talk about it from an operational angle, involving practical decisions. It should not have much to do with top management. Top management should be able to make use of data analytics in tactical decisions.”
In the big picture, strategy, new conquests.
“A few years ago, a major Finnish consultancy firm said that it doesn’t even try to sell tactical big data technologies to Finnish companies, as they were still preoccupied with using it on an operational level. I do believe change is on the way”, says Schildt.
“Avoiding a simple thought pattern of ‘show me the money’.”
Instead of short-term gain, companies would look to big data as a tool for long-term decisions.
“Of course the investment climate may put companies off the idea of investing many years into the future”, Schildt admits.
But he does promise good things to those daring companies that invest in tactical data mining.
“Data analytics can be divided into first, second, and third generation big data. The first generation optimized routine decisions to create cost savings. Second generation big data enabled completely new types of business models, such as Facebook and Google advertising. Or Trip Advisor, which centers its business on traveler reviews. Big data enabled seeing and building completely novel business types.”
And the third?
“The third generation moves from numbers to cracking text”, says Schildt.
“Cracking text data can for instance involve real-time analysis of Twitter data, and a possibility to create concept maps of the company and competitors. How do people talk about a competitor’s products? What tone do they use when discussing your products?”
Big data generated from text is even more challenging than numerical data, as numbers rarely involve hidden meanings.
“Numbers are easy, text really tricky”, says Schildt. “Algorithms aren’t particularly good at detecting irony.”
True: numbers are clear on whether results are good or bad, whereas it is not immediately apparent whether a customer’s comment “Good job, British Airways” posted on social media is meant as praise or irony, in other words criticism.
“However, text analysis does offer huge potential, especially when thinking about all the new, free text data masses continuously available. Data cannot be used at face value, but data analytics tools uncover a great deal.”
The faster pace at which big data is gathered and diversifying data forms – in addition to numbers and text, the world is flooding with video material captured by individuals on their cellphones – lead to figures that are even harder to fathom.
As mentioned earlier, the current estimate of data gathered around the world each day is 2.5 quintillion bytes, and it is estimated that 40 zettabytes of data will be created by 2020. To clarify: four years ago, the entire World Wide Web totaled 500 exabytes in size – i.e. 5 billion gigabytes, or half a zettabyte.
In other words, the 40 zettabytes looming in four years’ time is the equivalent of 400 billion gigabytes.
As the amount of data grows, also data management tools become more widespread. A great deal of what was pioneering data mining a decade ago is now easily and freely available to anyone.
“The main change hasn’t been in what you can do with data, but in the price of data mining tumbling down”, Professor Schildt assesses.
“An airline recently announced that it was about to acquire data warehousing solutions from a specialist company at a value of EUR 3 million, but a few years down the line chose an equivalent cloud-based service for EUR 200,000. Highly sophisticated data mining can now be carried out with software that is completely free.”
In other words, anyone can mine data – it is more a question of who knows the questions to ask the data, and comes up with the best or most optimal interpretations. Big players in the field have already averted their gaze from simple number crushing to advanced text analysis.
“IBM is one of the companies to invest in text data. Prices will be coming down and a great deal of open source services will be available also in that area”, Schildt predicts.
The professor encourages companies to look beyond short-term financial optimization, and aim for wisdom for strategic questions. He also advises to make use of development trends in data mining. Processing numerical data is cheap and sometimes even free, and good to use where it makes practical sense (“analytics is a quick method for improving people’s bad decisions”).
“Polarization will take place. Basics that don’t take great understanding will be completely automated in areas with huge amounts of data that people simply cannot understand. Machines will do the job more efficiently and cost-effectively. But as data automation spreads, also a need for genuine understanding and expertise becomes more prevalent.”
In-depth utilization of big data – “data-driven decision-making”, nevertheless led by visionary individuals – can result in completely new innovation. Schildt mentions online bookstore Amazon as an example, which began to put both customer purchase and search data to use. Users were after all revealing a great deal about their needs even without buying!
“This is where things should be heading. People need training in it. As business processes create big data, companies need innovative managers and other professionals who are able to systematically examine that data and identify ways to improve how the company operates and even what its goals are.”
“Big data is over-hyped.”
Dr. Jussi Keppo’s words startle – especially, as he utters them during his lecture entitled Leading with Business Analytics at Aalto Executive Summit.
Finnish-born Keppo is an Associate Professor who teaches risk management and business analytics at the National University of Singapore. He thinks there is too much hype around data masses – at least as far as size is concerned.
“Good data is the main thing. It’s better to have good data than big data. A company that uses the right information the right way makes the right decisions. How to find good, suitable data is a more difficult question.”
Here Keppo thinks along the same lines as Schildt:
“It’s so easy to acquire big data these days, but whether you can use it is a totally different matter.”
Keppo sees the generational gap as the biggest bottleneck in using analytics. Boardrooms continue in the hands of baby-boomers, decision-making ignoring those with genuine knowledge about data analytics.
“At the moment, the top executive level is represented by our generation and older. We think we know our stuff, and we can hold on to power by keeping decisions in our own hands. But you have to be dynamic in today’s world, rather than thinking that some executive board makes all the decisions. They are not even aware of all the data and analytics opportunities that exist lower down.”
“An organization doesn’t necessarily have to be flat, but in order to fully utilize data analytics in today’s dynamic competitive landscape, decisions need to be made on every level.”
At a coal mine, you need to listen to those working underground, and in extracting oil, employees with experience of life on an oil rig need to be heard. The same goes for data analytics: more decisions should be left to those with the best knowledge of the data and analytics. Many times these employees know the current situation better than top executives because they have the data and skills to analyze that.
Or at least decision-making should involve listening to those who understand the gathered data.
Big data requires a new mindset from managers. To grasp this, let’s return to Professor Schildt, who has spent a great deal of time examining the theme. Last summer, Schildt drew attention with his article on a “culture of secrecy” that still prevails in companies.
Although good reasons like data leak risks and privacy protection exist, usually the reasons for secrecy are selfish. Knowledge is power, and power makes life easier. In his time, Italian Renaissance philosopher and statesman Niccolo Machiavelli advised leaders to conceal the reasons for decisions in order to avoid their wisdom being questioned. The more information spreads, the more leaders need to justify their decisions”, Schildt wrote in his article for financial online publication Taloussanomat.
He sums up the called-for change in one word: “Transparency.”
A data overload demands transparency, also internally. Managers can no longer hide behind the data they govern. The old saying “knowledge is power” is just that, an old saying.
“This requires a major cultural shift in a lot of companies, which bosses may see as a threat. Traditionally, data analytics tools, for instance, were only available to a small circle in a company. A thinking prevailed that only top executives had the need and the ability to analyze and use data. Nowadays, tools and data need to be shared with everyone. Everyone on every level needs to be able to use it.”
A cultural shift means both transparency and trust. The principle is in fact logical: in reality, there are less secrets, so confidential data ending up in the wrong hands poses a smaller risk. However, your rival knowing how to process its own or generally available data in a smarter way than you is a real threat.
All competence on all levels needs to be harnessed into this competition.
“Of course analytics involves major ethical questions”, Professor Schildt admits.
“Customer and patient information require high ethical standards on data processing. It’s also important to remember that numbers aren’t reality, they just describe it.”
But even in problematic ethical questions, personnel should be approached with confidentiality, trusting that your competent employees know what to do with the data.
“Some American startups of course do first and ask later, which has its problems, too. When you head technology first and see what you can do, you wind up in problems.”
Schildt intentionally chooses the world “capability”. It is a future competitive advantage. An understanding of ethical perspectives is necessary for harnessing data wisely and efficiently in decision-making and day-to-day business.
An alternative recipe for success no longer exists.
Professor Schildt mentions two competing grocery retail chains, which practically shared the domestic market. One of these made use of its own big data, customer information on each purchase from loyalty cards, in its decision-making and business development. The other chain took its time, as it feared how customers would react if they used the data.
The faster chain had taken its lessons from the U.S, where for instance the transformations of Walmart and Target are seen as success stories in big data. Walmart was the first to offer customer data to product manufacturers. This way, information on product performance in stores was able to boost product development processes.
In the case of the two grocery store chains, it is quite obvious how the story ends. The chain that was slow on the uptake has conquered its fears by now.
The rivals are now engaged in a frantic data mining duel.
This is one of the long form articles in Aalto Leaders' Insight - magazine Vol. 5. The whole magazine can be read in Issuu: