Big Data in small retail?

We seem to be on the cusp of a major breakthrough that will bring big data to regular folk. So far, it has only been used by giants: the Obama 2012 presidential campaign is one of the best examples of widespread and highly effective uses of big data. But what about regular run-of-the-mill businesses? For years now experts have emphasized how much data regular storeowners collect without even realizing, and how much they could benefit from analyzing it. Why aren’t they doing it already?

By Beata Socha

Retailers have long bemoaned the so-called “showrooming effect” which means that their clients only come by the store to have a look around and then make all of their purchases in online stores, which offer all the same goods but cheaper. First, they complained about the cost of maintaining stores and paying rent. Then the industry discovered that if you efficiently combine your bricks-and-mortar outlet with a slick online platform, rent and storage costs can be offset by the online sales component. So-called omnichannel retail (which includes the ever-growing “click-and-collect” concept) is the buzzword of the retail industry these days.

However, retailers kept complaining. They claimed with utmost conviction that no matter how convenient and user-friendly online shopping gets, there is no impulse buying in e-commerce. “In a regular store you come and buy a dress, and then you couple it up with a scarf you see and then on your way out you grab a purse and buy the whole set. You don’t have that in online stores,” you could hear at a retail conference.

But things are about to change. If you collect information on what your clients buy and what ensembles they create, you can easily create the “wow, that goes great together” effect in your web store. It’s just like high-end personal shopping services: once you get to know your client, you can play to his or her tastes and preferences. With big data and analytics, you can do all that on an unlimited scale, and add even more: you can profile your clients and find similarities between them. For example, if client A bought a T-shirt with a pair of jeans and then next week returned to buy a striped red dress, perhaps client B, who’s looking at the same T-shirt might appreciate the dress as well. Zalando is one example of how utilizing suggestions from other clients can increase sales. “Big players are already using big data. Small entrepreneurs are at the very beginning of their journey,” said Artur Pająk, senior product manager at Huawei Polska, at a recent Big Data Congress.

Another typical use of big data – analyzing client movement around a store to manage the store’s display areas and goods placement – seems ideally suited for storeowners, big or small. “SMEs are more flexible [than big corporations] and can act on the data more quickly,” said Michał Niemczycki, deputy CEO at FreeBee. So why are blue chips still the only ones making use of big data?

Sunk costs

According to Niemczycki, despite small businesses’ flexibility, the practical implementation of big data in SMEs is virtually impossible. “Small firms lack the time, the resources, and the strategic understanding of the data.” The costs can indeed be overwhelming for a small firm if it wants to develop its own big data solution. Software development alone costs some PLN 70,000-PLN 100,000, and you do it without really knowing what the end result will be. Then there are license costs: the Hadoop platform goes for $10,000 per Terabyte. Then servers: a Terabyte RAM 64-core processor is an expenditure of roughly PLN 100,000-PLN 150,000, employing a data scientist for a year – another PLN 150,000. “We are already hitting the half-a-million mark, and we are talking about SMEs. That’s not a small challenge for them. Not all of them can afford it,” said Dariusz Piekarski, manging director and board member at SARE.

Clouds getting cheaper

Naturally, entrepreneurs will be looking for cheaper solutions that will allow them to analyze data about their clients. The first cost cut they can implement is in infrastructure. “Uber doesn’t have its own taxis and yet it operates a lot of cars. The same goes for cloud-based big data,” Andrzej Sieradz, CEE Enterprise Architect director at Microsoft’s Poland office, commented pointedly. Pająk also believes that the costs of such an endeavor can be effectively minimized. “You don’t need to invest in computing power – you can buy it as a service,” he said. “Cloud solutions will likely be the most popular way of using big data, because they’re cheap and flexible – you can simply unsubscribe if you no longer need it,” he added. Of course, there’s the small but important aspect of data security. Cloud solutions may be cheaper, but unless a company is 100 percent sure its data is safe, it will still prefer to keep it all in-house. Most companies also have no idea if the data they are collecting can be used for analytics. As much as 58 percent surveyed by consultancy KPMG admitted to having difficulties determining whether the data they have collected is reliable. “Companies are finding it harder and harder to make sense of the digital data deluge. Meanwhile, the amount of data will only grow larger in the next few years,” said Piotr Prajsnar, CEO of Cloud Technologies.

Human factor

If not the cost, then what is still getting in the way? According to Microsoft’s Sieradz, three things are necessary for big data to make a breakthrough in SMEs: cloud-based solutions that not only analyze data, but also allow to make predictions; technology allowing easy transfer and collation of different sources of data; while the third element is “understanding the data.” The human aspect, mainly the need for highly competent engineers capable of using big data technology, has been underscored by many. After all, if you don’t ask the right questions, even with the best state-of-the-art technology, the answers will be flawed. “Servers are just a fraction of the cost. The real cost is people, particularly competent people,” said Marek Niezgódka, PhD, head of the Interdisciplinary Centre for Mathematical and Computational Modelling, University of Warsaw. “The term ‘data scientist’ is applied too broadly,” he added. There aren’t all that many who fit the bill.

Luckily, as a head of an SME, you don’t necessarily need to hire your own data scientist to utilize the data you have collected. “The idea, the analytical element, and the most important thing about big data, can already be bought,” said Pająk.

Leap of faith

[caption id="attachment_98467" align="alignright" width="300"] Image: Shutterstock[/caption]

But even if we had all the right experts who could ask all the right questions and make accurate models, who’s to say a typical CEO would want to act on their advice? “In most SMEs the CEO runs his company based on his experience and intuition. He is confident in the know-how he has accumulated over the years and is reluctant to make decisions based on pure data,” said Niemczycki. “We’ve experienced that first-hand. We offered our clients very specific data on which they could act: traffic data, who comes into their stores and when. Based on the data they could modify their loyalty programs, e.g. change the rewards etc. But very few companies did it,” he explained. It seems that without a serious shift in how the C-suite views big data, and whether they are capable of trusting it, all the efforts of the IT biz will go unrewarded. “You need to trust the data. You can’t keep thinking: ‘I know better,’” urged Sieradz. Hope seems to lie with the youngest generation in the job market. “Millennials are the first generation that seems comfortable basing all their decisions on data and not on intuition,” explained Piotr Adamczyk, CIO of Work Service. “They no longer want to ‘slide’ on intuition,” he added.

‘Plug and play’

Apart from insufficient competences, the biggest barrier in bringing big data solutions to SMEs is, according to Piekarski, the fact that IT providers don’t offer a finished ready-to-use product, simply because they don’t have the necessary know-how of the business whose data they’re analyzing. “Instead, they want to learn by adjusting big data technology to our business, and that’s a major cost,” he explained. We still lack self-contained, off-the-shelf products that SMEs could just “plug and play.” The PC ushered in a new era of data handling over 40 years ago, a thing that was earlier reserved only for the biggest companies, where a host of people worked on closet-sized computers. The same happened when the iPod pulled the rug out from underneath the portable music device industry. Many would argue that the biggest advantage it had over the competition was not the size and look, and not its range of capabilities, but simplicity. “SMEs will be able to use big data once these systems become more automated: once we implement artificial intelligence and offer final conclusions instead of data points,” said Niemczycki. It seems that whoever figures out how to turn big data into “easy data,” will make a killing in the SME world.