The Faculty of Business and Economics
How can Big Data impact on store location selection process in
the consumer goods industry?
Master Thesis
M.Sc. Business Administration – International Management Track
Supervisor: Dr. Markus Paukku
Student: Petteri Max Rickhard Salmi
Student ID: 11186615
Submission Date: 24th March, 2017 Final Version Word Count: 18,842
Acknowledgements
I would like to thank and recognize Dr. Markus Paukku under whose supervision this research was conducted and who supported me through out the process. A special acknowledgement goes to all the individuals from the companies that provided time from their busy daily schedules to help in participating in this research.
Statement of Originality
This document is written by student Petteri Salmi who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating
it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Due to the emergence of Big Data as a phenomenon and much discussion around the actual value of Big Data, this paper aims to find out if and how the value can be gained in the retail store location selection process in consumer goods industry. How this influences the existing theory business world as Big Data changes the business environment bringing new opportunities and challenges. Through a parallel multiple case study of five different cases the results support that companies in Finland are using Big Data in their location selection process doing this mainly internally while showing some signs of wanting to outsource. In the location selection process, companies do not rely on competitors location as a main factor in their own store location selection process, but instead use other factors that should be added into future theories and models of the location selection process. Overall the companies have a view that Big Data can help their companies turn their store location selection process into a competitive advantage saving them time and money through making fewer mistakes in their location selection process.
Key words: Big Data, data-driven decision-making, location selection process, competitive advantage
Table of Contents
1. Introduction ... 6 2. Literature Review ... 8 2.1. Big Data ... 8 2.1.1. Open Data ... 10 2.1.2. Closed Data ... 112.2. Data-Driven Decision Making ... 12
2.2.1. Location selection process ... 17
3. Theoretical Framework ... 21 4. Research Design ... 25 5. Findings ... 30 5.1. Data collection ... 30 5.2. Data analysis ... 30 5.3. Findings and Analysis ... 31 5.3.1. Case One ... 31 5.3.2. Case Two ... 38 5.3.3. Case Three ... 44 5.3.4. Case Four ... 49 5.3.5. Case Five ... 56 6. Discussion ... 63 6.1. Cross Case ... 63 6.1.1. Proposition Validation ... 70 6.2. Managerial Implications ... 70 6.3. Limitations ... 71 6.4. Future Research ... 72 7. Conclusion ... 73 8. References ... 75 9. Appendices ... 81
Table of Figures
Figure 1 Table of Participants ... 30 Figure 2 Case 1 Importance of factors used in location selection process ... 34 Figure 3 Case 1 Level of data use in decision-making ... 35 Figure 4 Case 2 Importance of factors used in location selection process ... 41 Figure 5 Case 2 Level of data use in decision-making ... 42 Figure 6 Case 3 Importance of factors used in location selection process ... 47 Figure 7 Case 3 Level of data use in decision-making ... 47 Figure 8 Case 4 Importance of factors used in location selection process ... 52 Figure 9 Case 4 Level of data use in decision-making ... 54 Figure 10 Case 5 Importance of factors used in location selection process ... 59 Figure 11 Case 5 level of data use in decision-making ... 60 Figure 12 Main factor limiting Big Data use in location selection process ... 64 Figure 13 Avg. response of factors used in location selection process ... 65 Figure 14 Each response on the scale of 1 to 10 ... 66 Figure 15 Comparison of answers ... 67 Figure 16 Avg. vs. Ideal level of data used in decision-making ... 68 Figure 17 Proposition Validation table ... 701. Introduction
Merrified (2000) and Wood (2000) wrote about the idea that the Internet is a phenomenon that would make past business models outdated and that the buzzword ‘business model’ was associated with the new economy of the Internet up until the fall in early 2000. Now this raises a question, is Big Data merely a buzzword of our time which is soon to pass and lose relevancy or is it something that can be used to create value in different business processes. Large amounts of data has been used by retail firms in the store location selection process already in 1999 Grewal, Levy, Mehrotra, and Sharma, (1999). The value of information is high, information revolution was claimed to give companies a competitive advantage over their rivals (Porter and Millar 1985). Davenport and Harris (2007) take the debate onto Big Data claiming it can yield a competitive advantage for companies.
This research is important for both the academic world as well as professional world because Big Data is a relatively new phenomenon and the influence of it has not yet been studied extensively in the different business processes. With the aim to find out how Big Data is used or can be used to improve the location selection process by using data-driven decision-making thus be able to make location selection a competitive advantage for the firm. For the academic side, this research provides insight into how the location selection process has changed in terms of what are the most important factors to be considered in the process and what possible additions should be considered when making theoretical assumptions and models for the location selection process. Literature review of previous academic literature on the topic is outlined in the beginning. Followed by theoretical framework and methodology. In the end the findings and results are discussed in the light on the propositions for this research.
In this study five companies operating in Finland are analyzed in their current decision-making process in regard to the location selection process of their own physical stores. This seeks to find out the factors that currently guide companies in making these decisions and
how Big Data is used in the location selection process. Moreover, if Big Data can be used in the store location selection process to create value by making this process a competitive advantage for a firm. A parallel multiple case study method is used to study five cases individually through data collection via semi-structured interviews. This means that each case is studied individually not in a predetermined order but the interviews are taking place when possible, so that not one case has to be fully done before the next can start. As a qualitative case study, this study opens up new areas of research on the topic of location selection process. Qualitative research via case study approach is used, as this is a great method for studying phenomenon answering questions of “how” and “why”. (Yin 2013)
Supporting the aim of the study, the results show companies’ best practices as well as the future direction of what companies want to use in the location selection process to discover if Big Data use can yield a competitive advantage for a firm in the location selection process. The research question in this study is: How can Big Data impact on store location selection
process in the consumer goods industry? The study aims to answer if companies are using
Big Data in the location selection process and how can Big Data impact location selection process of companies by gaining a competitive advantage through this process. What kind of data can be used in the process and how does it help as well as giving light on possible future insights of Big Data use in location selection process.
2. Literature Review
2.1. Big DataThere are many working definitions that can be applied to Big Data but most clearly explaining the concept of Big Data the following definitions apply.
“Big Data generally refers to data that exceeds the typical storage, processing, and computing capacity of conventional databases and data analysis techniques. As a resource, Big Data requires tools and methods that can be applied to analyze and extract patterns from large-scale data” (Najafabadi, Villanustre, Khoshgoftaar, Seliya, Wald and Muharemagic
2015, p.6).
This means that it is data of enormous size that exists because of multiple variables in many different forms that are collected from different contact points including people and devices today, making this both wide and tall data. Big data analytics is taking this data, grouping and mapping it into categories in order to find anomalies and patterns in the masses. According to, Bughin (2016) the amount of data creation will continue to grow at a 40%-60% per year. This means that while we are entering an era of Big Data, this is still the beginning of what is to come. Meaning that Big Data will continue to grow even larger data over the following years. Big data can be looked at as the 3 V’s as described by Laney (2001) volume, velocity and variety. Volume is the size and amount of the data. As an example according to Manyika, Chui, Brown, Bughin, Dobbs, Roxburgh and Byers (2011) Tesco generates over 1.5 billion new items of data every month and Wall-Marts’ warehouse included 2.5 petabytes of data in 2011. Velocity is the speed at which data is being created. According to Manyika et al. (2011) retailers are able to track consumer behavior nearly real time due to click and purchase behavior online. Variety refers to the complexity of the data including multiple variables and dimensions creating data in different forms. Different type of data comes from different types
before. A case of IKEA shows how increase in IKEAs online traffic was linked with in store sales with bad daily weather in Norway. (MediaCom 2014) This shows that data comes in large variety however the usefulness of all data is debatable. There are also two other V’s that have been adopted in the Big Data definition, value and veracity. According to White (2012) Value means that Big Data can create benefits for companies by creating economic value in how it is used. As by IBM (2012) Premier Healthcare Alliance was able to improve healthcare services such as patient safety and identify best treatments while saving costs due to the use of Big Data and informatics in their system. Veracity means that some data might not be up to the quality and correct to what is needed when the data is grouped with all the other data. In this case data analytics might provide wrong matching results due to the ‘incorrect’ data input in the system. According to Davenport, Barth, and Bean, (2012) eBay had a massive problem with their data as the data collected got replicated in large quantities all over their systems, this became hard to work with and they had to create a new working system for their data hub.
White (2012) brings up the importance of filtering and human factor in this data so that it can be controlled up to some extent so that wrongful results can be avoided. To give an idea of the volume needed in order to get a ‘Big Data’ stamp according to Manyika et al. (2011 pg.1) “
“Big data” refers to datasets whose size is beyond the ability of typical database software tools to capture, store, manage, and analyze.“ The size of Big Data is so large that companies
with no internal capabilities and resources are not able to work with it. Big Data requires massive computing power, hardware, capabilities and resources from companies in order to be able to work with it. Therefore Big Data can be defined using the 5 V’s as large amounts of relevant data with multiple variables that is created and analyzed at a fast pace and later grasped to gain economic benefits from the data. The phenomenon really needs to be understood at all levels in order to be able to benefit from it. In the next part a further
distinction in Big Data is made between publicly available data and privately achieved data. This can potentially mean the difference between being able to use it hence publicly available and not having the funds and means to collect and harvest the closed data.
2.1.1. Open Data
According to Murray-Rust (2008) open data is data that can be published or re-used without permission barriers or fee to gain access to the data. While Murray-Rust (2008) mainly concentrates on the scientific data, this becomes important to note as much of management and international theory is based on scientific research. Murray-Rust (2008) concludes that open data would be very beneficial for the scientific work if the data can be mined and analyzed in a manner that can aggregate data. This shows the value of large data sets even Big Data providing valuable insight for scientific work. When looking at the open data from another point of view, there are multiple databases and data sets that can be accessed freely. According to Huijboom and Van Den Broek (2011) an open data strategy is adopted by ever more western countries and governments as an act towards transparency. This could mean that the availability of open data should increase as governments provide and publish more data openly. Gurstein (2011) covers in his work that while open data is valuable and empowering, open data becomes value adding and empowering only to those parties that have the knowledge, skills and basic infrastructure to deal with the data. According to Benjamin, Bhuvaneswari and Manjunatha (2007) digitization of land records meaning opening land ownership records to public use resulted in companies targeting so called poor owners with their purchase offers in the mindset that they could purchase land for a bargain. This can be seen as information that can be used by companies when deciding their choice of location for setting up their stores in. Arribas-Bel (2014) noted an example of large-scale data about transit by subway (hourly volumes, stations used to get out and enter among other data) that was made public by the New York City in effort to provide possible use for companies for
free. This sort of data can be crucial for a company when deciding on the location of the store. Boyd and Crawford (2011) explain that for example an open source for data is social media such as Twitter apart from the accounts that have been made private. This opens a whole world of access to data for free if they have the knowledge skills and system to draw the data with and work with the data.
2.1.2. Closed Data
Cobden, Black, Gibbins, Carr, and Shadbolt (2011, pp.1) define closed data as “access to and
use of the data is subject to legal or technical restrictions which go beyond attribution and share-alike obligations”. This can also be defined as data to which access and usage can be
purchased/ granted or gained via means of payment. According to Andrejevic (2014) being able to use Big Data requires the expertise, technology, hard- and software and access to the data. Hashem, Yaqoob, Anuar, Mokhtar, Gani and Khan (2015) bring up the debate to cloud computing as a step towards easier adoption of Big Data as not such massive physical storage capabilities would be needed. However, they state that this technology is not at the convenient usage level yet and developments need to occur in the area before this could be easily adopted for use (Hashem et al. 2015).
Andrejevic (2014) notes an important factor, who is (the legal owner) thus owning the rights to the data. According to Goodchild (2007) people combined with mobile devices are effectively becoming sensors by themselves that create networks of sensory data emitted by the devices in multiple ways. For example the data emitted is geographical location data through the use of internal Global Positioning System (GPS) and Geographic information system (GIS) which captures, shared and displays data related to location and positions on Earth. Arribas-Bel (2014, pp. 4) says, “Intimate aspects of humans are being stored online.” Whether we as humans like it or not this comes from social media posts, feelings, emotions
(2007) the privacy paradox is an occurring phenomena meaning that people disclose willingly much more information about their personal preferences and tracking on their devices that they claim to want to. This brings up a limitation in closed data, as the boundaries need to be clear as what data can be used. Even if the data can be collected and accessed, does not mean the data can necessarily be used. This itself is another topic for a research paper but this is brought up merely to explain that with closed data there are certain obstacles and limitations that might implicate to use of it.
2.2. Data-Driven Decision Making
Found by Bakhshi, Bravo-Biosca and Mateos-Garcia (2014) that firms using data from online both by collecting and analyzing were more productive than the non users. Gantz and Reinsel (2012) state that in 2012 there was 23% of the data available that would be useful if it was tagged and analyzed. This means that while the data collection is at record numbers, the data is actually useful for companies in different industries only if it can be tagged and analyzed appropriately. The productivity of companies using Big Data can also be linked to data-driven decision-making. Brynjolfsson, Hitt, and Kim (2011 pg.1) say, data-driven decision-making can be defined as “decisions that are based on data- and business analytics”. Brynjolfsson et al. (2011) found in their extensive survey that companies who have adopted data-driven decision-making process in their business practices and information technologies were in total 5-6% more successful in their output and productivity. Providing a clear indication that data-driven decision-making is linked to business performance in terms of output and productivity when adopted. This is relevant as in this research the aim is to find out if in fact Big Data use can provide value for the company in terms of competitive advantage in the location selection process by an improved -process. The importance of data-driven decision-making process with the research question is that companies who want to use or already use Big Data in order to gain competitive advantage do they have or need to have a data-driven corporate culture in
order for them to be able to extract the value from Big Data. With enormous real-time data available for firms, there is no reason why this cannot be turned into a competitive advantage (Burrus 2015). Furthermore in order to harvest and utilize data in the location selection process a company must be able to work with the data. And is it essential that a company has a data-driven culture in order to be able to extract value from the data. Also can Big Data be taken on if the current data efficiency is not at its peak in a firm can be arguable. Alharthi, Krotov and Bowman (2017) also emphasize as culture being of the limitations to Big Data adoption explaining that the corporate culture needs to be adjusted so that it can handle a data heavy Big Data operational style and use this for value through data-driven decision-making. White (2015, pg.1) said,
“Derived from a sample of 1,650 businesses that responded to 36 survey questions, the Information Value Index gives businesses a score from 0 to 100, with 100 being the best use of data possible. This index evaluates a company's general awareness and understanding of the importance of data, how aligned the company is with data driven goals, the skills and tools used to gain value from data and overall benefits the company has gained from tapping into data. Mid-market companies earned an average score of 48.8, while enterprise businesses earned an average score of 52.6; combined, the overall score for all companies surveyed came in a just over 50.”
This brings a valid thought about data use currently in firms and opens up the idea of data efficiency with data currently available. Is data efficiency a prerequisite to entering Big Data or is Big Data adoption useless in such case? According to Ross, Beath, and Quaadgras (2013) the biggest problem that investments in Big Data fail is because companies don’t know how to take advantage of the data they already have. They also studied a financial institution to find out best location for their ATMs but found out that consulting firms worldwide have already built models that help in determining ATM locations Ross et al.
(2013). This shows the importance of understanding if data currently available is being used in an efficient manner. While store location selection process is considered a data heavy process, there is supporting research arguing that data in decision-making should be only used as evidence for decisions made by intuition and experience (Jones and Mock 1984). Also found by Hernandez and Bennison (2000) that many retailers even prefer rather intuitive decision-making process compared to data-driven decision-making process.
Davenport and Harris (2007 pg.9) say that “Unique geographical advantage doesn’t matter in
global competition, and protective regulation is largely gone. Proprietarily technologies are rapidly copied and break through innovation in products or services seems increasingly difficult to achieve.”
According to Davenport and Harris (2007) competition is getting fiercer globally and this is why analytics has to be emphasized on in order to take maximum output and efficiency out of the business processes as analytics can support most business processes. The focus of store location selection process is interesting to see if this is something that with Big Data input can create a competitive advantage for a firm by improving the process by means of better data use. The advantage of using Big Data for example according to Watson (2014) Starbucks used Big Data when coming up with a new coffee product. They monitored social media channels for comments about the new coffee and found that the flavor was in general good but the price was found too high. They lowered the price and later that day all negative comments in the social media had disappeared (Watson 2014). Data-driven decision-making is debated to change the influence of theory as theories are brought up by scientists, hypothesized and tested. But in the era of data-driven decision-making theory is no longer be needed as numbers and data speaks for itself making data-driven decision-making the most efficient way to proceed (Anderson 2008). However, McAfee, Brynjolfsson, Davenport, Patil and Barton (2012) argue that even though Big Data can be very useful in taking data-driven
decision-making to the next level thus making better business decisions in the future. McAfee et al. (2012 pg. 68) say that, “Big data’s power does not erase the need for vision or
human insight”. But the leadership skills to run the team working with the data remain
exceptionally important. McAfee et al. (2012) outline 5 management challenges that are evident when dealing with Big Data. The challenges are 1) Leadership meaning that teams need to be able to ask the right questions and define clear goals. 2) Talent management, with cheaper data the human resources to handle the data become more valuable as they are new skills and knowledge working with Big Data. 3) Technology, having the right technology to work with the data and be able to make use of the data by analyzing it. 4) Decision making meaning cross-functional teamwork working with both information and relevant decisions in the same direction. 5) Company culture meaning changing the way things are done, working with concrete facts e.g. numbers and data to arrive at conclusions instead of a gut feeling. Big Data speeds up decision-making process. This model and theory behind decision-making requiring human factor is linked to the behavioral theory by Henry Mintzberg who strongly supports and researches about the importance of human factor in strategic planning. He believes that the strategic planning process requires vision of the manager and others around meaning this can rise from any level as well as hard data as evidence (Mintzberg 1994). Adopting Big Data means that a company must take on a data-driven decision-making process in order to get value from it (Virkki 2015). Anni Ronkainen (CFO, Kesko) says, “As
a manger I feel free because I can make strong decisions based on data and not ‘I feel like’ gut decisions” (Vuokola 2016 pg.5).
A part of the location selection process is strategic planning, one definition of effective strategic planning is outlined by Dyson and Foster (1980 pg. 163) as “a system that sets goals
and achieves them within a specified time.” Doing this as the process of using information
can be also linked to strategic planning. An early link of the information revolution effecting competition made by Porter and Millar (1985) claiming that the information revolution can give a competitive advantage to companies thus outperforming their rivals. Doing this by creating value in the processes by adding the new information in to the process. In a research study executed by IBM Institute for Business Value (2012) a group of 1144 business and IT professionals were interviewed. From the report results show that 63% of the respondents in 2012 said that they are creating and using Big Data and data analytics to competitive advantage for the firm. The same number in a study also conducted by IBM in 2010 was 37% from global respondents (IBM Institute for Business Value 2012). The increase in only 2 years is enormous meaning that companies are taking on Big Data and using it for their advantage. Therefore while the importance of information has been present in the creation of competitive advantage for a long time, recent results show that companies are taking action with the new availability of Big Data and capitalizing on the potential to create competitive advantage. According to Davenport and Harris (2007) in order for a company to gain a competitive advantage, the value is highest in the analytics, future modeling and data optimization level and not simply at the data gathering and better data input level. However one large issue with Big Data analytics found by Ross et al. (2013) is that Big Data analytics requires a specific evidence-based decision making culture within the firm. This might mean that a company must change their organization culture and this might be such a large change that even though improvement through data-driven decision-making process could be achieved the change might not be worth it. As an example of better decisions based on data.
“One retailer, for example, learned that it could increase profits substantially by extending the time items were on the floor before and after discounting. But implementing that change would have required a complete redesign of the supply chain, which the retailer was reluctant to undertake” (Ross et al. 2013 pg.90).
Thus while studies show improved performance through data-driven decision-making process, a clear line needs to be found at what level and what stage this is realistic for a firm to take on. Therefore concentrating on the process of location selection is legitimate as the scope of this research. An interesting statement by one of the leading Chief Digital Officers of Finland Katri Harra-Salonen, (CDO, Finnair) said, “Data cannot be a strategic competitive
advantage for a company without the consumer having transparent information about what the company does with the data” (Vuokola 2016 pg.24). Once again reminding that there are
multiple factors that influence Big Data use in a company however, this is potential for another research study.
2.2.1. Location selection process
In this part the focus is on the process of location selection of a new store. Already in 1999, Grewal et al. (1999) noted in their paper that large retail chains in the United States of America are using large amounts of data stored in their data warehouses to determine retail store locations. Companies aim to capitalize on competitive advantage in several ways, Runciman (1998) argues that geographic location is the best way to create competitive advantage as physical location in the market is tough to imitate to the exact same. General theories are discussed in short that have provided groundwork for location selection for a long time already on which more recent work is expanded on and focuses on specific factors. Although the main focus of this study is on a national level in this research, of course international application can be drawn from the same theories and applied to the cases. The first theory discussed is the Central Place Theory originally published by Walter Christaller in 1993 (Taylor, Hoyler and Verbruggen 2010). The Central Place Theory covers two important concepts, threshold which is the minimum population that is required to bring a good or service and make it available. Range is the distance consumers are willing to travel for a good or service (Mulligan 1984). This is very important as transportation routes are a
factor that is tested in the cases in terms of importance of it in the process since the theory supports it as very important in the location selection process. Taylor et al. (2010) evaluate their addition as central flow theory, which they claim as an addition to the central place theory. The difference is explained as in central place theory the place creates flows, and in central flow theory the flows create central places (Taylor et al. 2010). This can be applied to the location selection process of firms as it depends if they locate in central places already or concentrate in new places hence looking at migration patterns and future construction and housing projects when planning their location. It is obvious that both factors of the central place theory are important but do vary depending on what is the good or service that is being sold. The Central Place Theory explains that some services or goods are being sold (located) in specific settlements or central places depending on the factors mentioned above (Mulligan 1984). This research aims to find out if it is useful to involve Big Data in the location selection process as this could help set the store in a better place than just if a company places it in any central location. Companies locate in settlements close to each other because companies want to be close to their customers as the customers can purchase online now as Lauri Sipponen (CEO, Lidl) said in an interview, “Now you want to be close to the customer”
“Customer research shows the biggest reason for why consumers don’t shop at Lidl is because the stores are not on the routes people take” (Nupponen 2015 pg, 1).
Through careful analysis of the data, Starbucks is able to project foot traffic and average customer spend of a given location, therefore helping Starbucks to determine the economic viability of opening a store in that spot (Digital information and transformation 2015). Big Data use to provide foot traffic figures are highly used in the industry. Location intelligence can help retailers determine where their customers live. Retailers can use the location data about their customers to plan new store locations (Pitney Bowes). It is obvious that these both factors are important but do depend on what is the good or service that are being sold.
Another theory that is used to explain location selection is the Hotellings Law which in simplistic terms is a theory for competition and explains why competitive sides for example, firms set up in close proximity to each other in order to gain the maximum number of customers from a specified area in a linear market (Hotelling 1929). This theory encourages competition as two rivals are operating close to each other, the customers are left the choice of either one if location is the deciding factor. When more than two firms are present in a market the principle of local clustering takes place. This is when a new firm enters a market or an existing firm relocates, they tend to locate as close as possible to the existing firms in the market (Eaton and Lipsey 1975). According to Eaton and Lipsey (1975) the Hotellings Law is not a realistic application in a market with more than two firms, and an important factor is that customers are usually not indifferent and rational between the firms meaning they tend to lean over to one firm for other reasons. This is why the principle of local clustering is more relevant still concentrating on locating close to competition. However for the purposes of studying location selection the principle of local clustering may show useful results depending on the strategies of the company in the location selection process.
As a more recent input to build on Hotellings law also known as the principle of minimum differentiation there are more factors that need to be considered when making a location selection than merely competition Turhain, Aklin, and Zehir (2013) provide a classification of store location criteria. This criteria consists of 7 factors as follows: 1) Performance measures: as they found that better performance in a store due to its location is more likely to yield higher utility. 2) Population characteristics: this is knowing the structure of the population in and around the area, as well as the demographics of the population including but not limited to household size, purchasing behavior, travel time, density, political attitudes. 3) Economic factors: while this is partially included in population characteristics, some important economic factors should be included such as purchasing power, rentals and elasticity of rental
contract period according to Erbıyık, Özcan, and Karaboğa, (2012). 4) Competition: Turhain et al. (2013) propose that both direct and indirect competitors should be taken into consideration as well as a deep analysis into their facts and figures when deciding on a store location. 5) Saturation level: this is the index of retail saturation meaning what is the situation of current demand with the number current suppliers. 6) Store characteristics: This is subdivided into 3 sections. (1) Ease in accessibility (2) store-image attributes (atmosphere and assortment) and (3) physical costs of the store due to location (Turhain et al. 2013). Local store assortment can be a tactic that companies use to gain an advantage over their competitors by creating specific store assortment in local stores depending on the location of the store (Grewal et al. 1999). This brings another important factor in to the picture, to use both the store location and the store assortment together to gain a competitive advantage. As found by Runciman (1998) specific product assortment or differentiation can be used to create competitive advantage by companies. 7) Magnets: Specific crowd points such as government organizations, culture educational organizations, crowd points such as hospitals and markets, and vehicle maintenance points (Kuo, Chi, and Kao 2002). This model through the study by Turhain et al. (2013) provides a clear indication of multiple factors that can impact the location selection of a store. While concentrating on the competitors’ location out of the factors above, the aim now is to find out how Big Data can help companies create a competitive advantage in the store location selection process, and what are the main factors used in the process.
3. Theoretical Framework
Grewal et al. (1999) noted in their paper that large retail chains in the United States are using large amounts of data stored in their data warehouses to determine retail store locations. The data use in location selection process has increased over time and companies are using more and more data in order to select a specific location. According to Ross et al. (2013) the biggest problem that investments in Big Data fail is because companies do not know how to take advantage of the data they already have. According to Reno (2012) Big Data can be accessed from open sources such as social media unless the users are set as private accounts. Therefore while it could be retracted from open sources, only companies who can make the use of open data an effective value add (who have the capabilities and resources to use it) in their location selection process are able to do this.
Companies that have internal capabilities of working with data take an advantage of open data sources. Gurstein (2011) notes in his work that while open data is valuable and empowering, it becomes value adding only to those parties that have the knowledge, skills and basic infrastructure to deal with the data. Organizations must have the capabilities to leverage open data (Van Kuiken and Chui 2014). Therefore if a company does not know how to use the data they currently have, might indicate they lack resources and capabilities to take on Big Data (more data).
P= Proposition
P1) Companies are using mainly open source Big Data in their location selection process. P2) Companies want to outsource Big Data processes due to heavy requirements in internal capabilities and resources.
P3) Main reason why companies don’t use Big Data in their location selection process is the lack of internal capabilities and resources.
Competitor location has been an important factor showing reasoning for a specific location over many years. The principle of local clustering is when a new firm enters a market or an existing firm relocates, they tend to locate as close as possible to the existing firms in the market (Eaton and Lipsey 1975). The Hotellings law, the principle of local clustering also the Huff gravity model initially brought up by Huff (1964) explains store attractiveness comparable to competitors, under two key dimensions. 1) The size of the store and 2) the travel time to a store which is highly important when it comes down to the physical location of the store. Later brought up Saidani, Chu and Chen (2012) who also propose huff like models, where competition information is valued and where the profitability of the store is highly linked to the location of them in relative to the competitors’ location. These all consider competition highly important in the location selection. Also found by Redondo, Fernandez, Garcia and Ortigosa (2010) as explained in their paper that as company (A) locates a store in an area where a competitor (B) is already present, the company (B) follows and sets another store in the area thus pulling more customers away from the company (A) in the area. These explain the importance of competitors’ location in location selection process. These give reasoning for the next proposition in light of importance of competitor location in the location selection process to see if competitor location is a factor that is independent from external changes that Big Data brings.
P4) Competitors location still remains as the main determinant in the location selection process.
The discussion around decision-making process and how it composes of ‘art’ and ‘science’ meaning human intuition and data as evidence and basis for the decision-making. Hernandez
and Bennison (2000) found in their survey that many retailers rather rely on the “art” aspect covered as by human intuition rather than science in this case referred to as data in their approach to reach location decision. This also supports earlier findings by Jones and Mock (1984) and Mintzberg (1964) that state that data should be used in decision-making merely as evidence to back up decisions that have been made using human intuition and gut feeling through long experience indicating the importance of both data and human factors in decision-making. This is also supported by McAfee et al. (2012 pg. 68) they say that, “Big
data’s power does not erase the need for vision or human insight”. The leadership skills to
run the team responsible for the data, remains exceptionally important. Of course critique to this view is evident as more and more data is available in the business world and data validation techniques are ever more advanced. Anderson (2008) claims that data will overtake theory in business practices and decision making process and so that theory is not be needed in the processes as data drives decisions that produce better performance. Brynjolfsson et al. (2011) found in their extensive survey that companies who have adopted data-driven decision-making in their business practices and information technologies were in total 5-6% more successful in their output and productivity. Providing a clear indication that data-driven decision-making is linked to business performance when adopted. And being able to increase efficiency in picking a store location, can save a company time, money and yield with better locations.
Although data collection, processing and analyzing methods are becoming more advanced the idea that human intuition cannot be erased from the equation is strong in previous literature. The value of Big Data usage is also a topic with enormous importance when talking about data-driven decision-making. In the end it is about how Big Data can create value for the companies. Providing value by helping companies create a competitive advantage in their location selection process is the focus here. From the report IBM Institute for Business Value
(2012) results show that 63% of the respondents in 2012 said that they are creating and using Big Data and data analytics to competitive advantage for the firm. Even Porter and Millar (1985) and Davenport and Harris (2007) indicate that Big Data and information revolution can give a competitive advantage to companies. Thus this research focuses to find out if and how the competitive advantage can be gained in the location selection process with the use of Big Data.
P5) companies want to adopt a more data driven decision making process in their location selection, however they do not want a fully data driven process and keep human intuition as a factor in the end.
P6) Big Data can be used by companies to create a competitive advantage in the location selection process.
4. Research Design
This research aims to investigate how can Big Data influence a more efficient and data-driven decision-making process when dealing with the store location selection hence create a competitive advantage for the firm. The concentration of the study and results is in the consumer goods industry as this is closely related with very important retail business environment. This study looks at what has been done in terms of data use in the process as well as what data companies want to be able to incorporate into their location selection process. Essentially to see if Big Data can be used to create a competitive advantage for a firm in as early as the location selection process.
The unit level of analysis is what kind of data each company is using in the location selection process. This is a descriptive study, specifically used to allow understanding of the current situation in each case. Built on a deductive approach, this study uses existing theories as the basis on which propositions are built on and tested through the data found in the cases (Eisenhardt and Graebner 2007). A non-probability sampling method is used to select the cases as non-probability method is often used in a case study design (Wilson 2014). A purposive sampling method is used in both company and participant selection, which refers to the judgment of the researcher when selecting the cases as specific and unique for the study (Wilson 2014). Five companies are selected in order to get an understanding of the industry in Finland and to find the potential opportunities for companies. A wider selection of companies was not applicable in the time frame given for this research project but encouraged for future studies.
This is a comparative study of five cases analyzed individually and compared in a cross case analysis allowing for wider understanding of the situation and results. Three cases are consumer goods companies selected for their similar business environment and number of stores. One case is an advertising & infrastructure company, which is relevant to this case
because for this firm the location selection of their units is at the core of their business and can provide interesting insight that can be applied to the consumer goods industry. The last company is a consumer service company in the food sector and selected also for the companys high presence in Finland in establishing new locations continually. The selected companies show continual interest in establishing new store locations. This being evident in the interviews, this is why the topic is also relevant in the practical world.
A list of the companies and the individual participants from each company can be found in Figure 1 in the findings. The results give an overall view of how and what companies are doing currently with Big Data in the location selection process and what opportunities should be capitalized on in the future in the location selection process through Big Data.
This multiple case study is set in a parallel frame. This method reduces bias as the cases receive same treatment as the researcher does not apply learned knowledge only on to following cases that can happen in a sequential frame (De Vaus and de Vaus 2001). The study consists of semi-structured interviews and questionnaires. The participants are employees such as Chief Executive Officers’ (CEO)’s and retail store operations managers among other managers of the companies. Both methods are used due to time constraints and availability of participants. The study consists of qualitative research as this is often well associated with case study method (Yin 2013). In the light of construct validity case studies are argued to have a bias of confirming researchers preconceived notions (Ruddin 2006). However the outlines for each case are set at the same points attempting to make sure that the cases are executed in the same manner for example in terms of interviews. Also in the beginning of the interviews a definition of Big Data is addressed by the researcher in order to make sure that everyone has the same idea and understanding. This is visible in the appendix 2 in the interview outline. External validity is about the generalization of the results. In a descriptive case study usually a question of “how” is answered (Yin 2013). The use of multiple cases and
having a clearly stated objective of answering the ‘how’ question are important pieces in order for the research to be generalizable, of course caution should be used when generalizing from case studies. Reliability is the extent to which the data collection and analysis produce consistent findings in terms if another researcher replicates the study in the exact same manner (Yin 2013). All the material is recorded for the purposes of this study, as well as an outline of the questions asked is presented in the appendix 2 for the best possible replication of the study to confirm reliability. Of course the study took place in a continually changing environment in both academic and professional setting thus making it more difficult to replicate.
According to Creswell and Miller (2000) Data triangulation and investigator triangulation are limited due to time constraints and the fact that the research is ran by only one person.
Interviews cover the topic of retail store location selection process. What are the drivers and data used in the decision-making process in each specific case in the location selection process. Some questions require a numerical estimation on a scale from 1 to 10 from the respondents. This scale is selected as an accurate scale used to get an idea of what is the importance of a specific factor in the location selection process. This allows the respondent to indicate a relatively accurate level not leaving room for researchers influence as for example a scale from 1 to 5 might do. All interviews are recorded and transcribed and the transcripts can be accessed from the researcher. The interviews conducted took around 25 minutes each and each case consists of one to three respondents per company. With an original aim of completing all ten by interviews in the end eight interviews were conducted and two sets of answers were received through a questionnaire with the same set of questions in order to reduce bias however, every case consisted of at least one interview. Confidentiality and anonymity are addressed as an option for all companies to make sure that the study is conducted within ethical guidelines. All the data is recorded and the qualitative data is coded.
The chart with the coding can be found in appendix 1. Coding enables for easier analysis of data under similar ideas and be able to draw up to conclusions from the coded chunks (Miles, Huberman, and Saldana 2013).
The strengths of this method are that specific company analysis allows unique and detailed results from each firm that can be collected and compared. As by Yin (2013) case study method is a reliable method at answering questions of ‘how’ and ‘why’ which are specific to this study. In this method the advantage is to gain the point of view of the ones being studied (Pratt 2009). In-firm interviews allow for highly valued company specific knowledge. Also, it can be drawn that there was no corroboration between the respondents, even though the interviews per case were conducted during separate days, the results show much opposing views in many cases providing insight that there was no corroboration between the respondents which is something that should be taken into consideration in an interview method (Yin 2013). A qualitative case study is a good method to open up possible new areas of interest and research (George and Bennett 2005).
Some of the limitations that this study encounters are, Big Data being a current hot topic however, the lack of the studies completed in this area. Even though this is something the study aims to find out (the impact) the uncertainty of the future direction of Big Data is unknown. With the fast pace of change currently, the future application of the results is something that needs to be addressed with caution. With the short timeline, the phenomenon cannot be studied in full and this is why the study concentrates only on store location selection process, with few factors as measurements this may limit the results of the study as there are many other factors that can be measured which can be direction for future research. Since companies working with Big Data, valuable new insights are studied, one must understand that confidential information is not given out by the employees and this might reduce the width of potential in the results. A limitation of case studies noted by Seawright
and Gerring (2008) caution should be used when generalizing from case studies. The evidence is based on current actions of the location selection process as well as future needs/wants of the companies thus confidentiality in regard to possible secrecy has to be kept in mind.
5. Findings
5.1. Data collectionThe data used for this analysis is based on data collected through interviews and questionnaires, as not all participants were available for an interview on such a short time frame. In total 5 companies were selected, in the research overall 8 interviews were conducted and 2 set of answers were received through a questionnaire for the analysis due to lack of time. Overall chart of the respondents and companies can be seen below in the figure 1.
Company
Number of locations (stores) in Finland
(approx.) Position Name
Case 1 Fiskars 30
Retail performance
manager Johanna Järvenpää Director of retail
operations Finland Mikko Koponen
Case 2 Carlson 30 Manager Sales and Marketing Mika Heiskanen
CEO Kyösti Karhunen
Case 3 JCDecaux Above 6000 units
CEO Klaus Kuhanen Manager Anonymous (Q)* Marketing Manager
Sampo Koskinen (Q)*
Case 4 Finlayson 30 Retail Operations
Director Kaj Wasastjerna
Case 5 Koti Pizza 260 Manager Marko Fonsen
Sales Manager Markus Kaatranen
Figure 1 Table of Participants
Source: Author *(Q)= questionnaire
5.2. Data analysis
For the analysis, multiple codes are produced to gather key words and phrases as evidence either supporting or contradicting the propositions under the research question. This is a great method for aggregating bits and pieces, phrases and words into smaller groups of data from the overall data collected. Ideas represented by the codes help to gather and sum up
information in order to answer the research question using a qualitative method of data collection and analysis. Also weight of certain factors in the companies’ location selection process is given a value in each of the interview and these values are compared both in case as well as cross case scenario.
5.3. Findings and Analysis
5.3.1. Case One
Fiskars
Case one is a consumer goods company Fiskars. Fiskars was founded in 1649 and currently has over 8500 employees (Fiskars annual report 2016). Around 30 Iittala stores (A Fiskars subsidiary)(Iittala 2017) are currently located in Finland among others operating worldwide. Concentrating on the stores in Finland. Two persons from the Fiskars retail operations team were interviewed for this case. Full list of supporting statements can be seen in a chart in Appendix 1.
As research shown by Reno (2012), Manyika et al. (2011) and Gurstein (2011) indicate that power of Big Data can only be used if a company has internal resources and capabilities to take advantage of the data. These can be linked to what was found that Fiskars does use Big Data in their location selection process but only in very early stages and not in a systematic process but with a mixed method of internally and externally with majority of it worked with internally and coming from open source. However contradicted by the other interview the differences are clear. The higher ranked employee said, a majority concentration of open source Big Data used in their location selection process thus P1, is partially supported. Both respondents indicating Big Data being used internally especially the person higher ranked in
the hierarchy stating that most of their Big Data is being used rather internally than outsourcing, this contradicts P2.
“We are not very good at that but we are trying to get better and uhm we we have quite a lot of data existing” (Koponen, Fiskars)
“We use city information and data of big cities but not in a systematic way” (Jarvenpaa,
Fiskars)
“Mostly internally now” (Koponen, Fiskars)
“Yes, we do it both internally and externally.”(Jarvenpaa, Fiskars)
(Asked about open source data use) “Uhm no, I haven’t at least come across.” (Jarvenpaa, Fiskars)
The main reasons outlined why Fiskars is not using Big Data turned out to be the lack of capabilities and resources. This was heavily enforced by both of the interviewees with no discrepancy in the answers giving clear support for P3.
“The lack of capabilities has been earlier one thing that we haven’t had in house capabilities to do that we also haven’t had all the uhm resources nor the data available” (M.K Fiskars) “We don’t have the currently the resources” (Jarvenpaa, Fiskars)
When talking about the factors used in the location selection process currently and their importance, the answers were relatively similar in both cases. However, the importance of migration patterns is much higher for J. Jarvenpaa than M. Koponen opening up the idea of how and what data is understood at what levels of the company. The main factors used in the location selection process by the company are all in all traffic, competitors, area demographics, rental space, window size, distributors, entrance and customer flow. The most
important factors in this case out of the interviews turned out as own distribution and footfall in the area, which were different views in both interviews once again and very different in terms of their effect on the business.
“It’s our own distribution in Finland” (Koponen, Fiskars) “The footfall is the most important.” (Jarvenpaa, Fiskars)
Also noticed as shown in the literature review and theoretical framework, the location of competitors is a relatively important factor to the company when making a decision on location selection. Redondo et al. (2010) and Saidani et al. (2012) emphasizing on the importance of competitor location. As can be seen in figure 2 below, this is 3rd highest ranked out of the ones asked however this is also the lowest importance with another three factors ranked by J.Jarvenpaa. The answers of M. Koponen was not available for this factor leaving room for debate on this, but P4 currently as not supported. This is also as both interviews said that consumer data is more important for the company than competitor data. The fact that one interview showed a factor other than the ones shown below as the most important factor for them provides insight into future application of the factor into the location selection process.
Figure 2 Case 1 Importance of factors used in location selection process
Source: Author
Case 1 shows a clearly that the opinions on the current situation at the company are very different depending on the person interviewed. The person at the higher level in the corporate hierarchy has ranked current data usage at 5,5 points lower on the scale from 1 to 10 when asked about it. Also the current corporate culture in terms of how data-driven it is was ranked 3,5 points lower than his colleague who is ranked lower in the corporate hierarchy. While the employees both believed that an ideal level of data used in decision-making would be at 8,5 the differences open interesting information about the company maybe suggesting that the data knowledge is at the higher levels in the corporate hierarchy. If this is the case, ranking current efficiency in data use at a 2,5 indicates a clear inefficiency and this should be kept in mind when looking at the Big Data usage inside the company.
0 1 2 3 4 5 6 7 8 9 10 Im p or ta n ce o n sc al e fr om 1 to 1 0 Factor used in location selection process
Case 1
Koponen Mikko Järvinen JohannaFigure 3 Case 1 Level of data use in decision-making
Source: Author
“Its not really data driven, and that’s actually one thing I really wana change.” (Koponen,
Fiskars)
The ideal level ranked by both at 8,5 is still a fair distance from their current situation, which is ranked 3,5 and 7 by the respondents. The 8,5 indicates support for the research by McAfee et al. (2012); Hernandez and Bennison (2000) showing that a human factor is required in the decision making process and it cannot be all data driven. This is also supported in the comments by the respondents. This is clear support for the Proposition P5 in both of the interviewees.
“Yeah somebody to look at the I would see that in our business that’s still something that we need the kind of the human touch” (Koponen, Fiskars)
0 1 2 3 4 5 6 7 8 9 10 How data driven is the corporate culture at your Tirm? How much do you think data driven decision making should be as a part of decision making process? How efTicienct is your Tirm in using the data currently available? Sc ale from 1 to 10 Answer of each respondent
Case 1
Koponen Mikko Järvinen JohannaThe interviews show a clear direction Fiskars wants to head to as both interviewees indicated that data they would want to have available but do not currently use is customer traffic data but in more detail about how their customers move around a specific area and tracking their customers movement in a larger picture. And this is where they found an area in that Big Data could help in providing this kind of data for the company.
“Uhm well, of course the foot fall where people they are passing by and the routes and the age and the income all the factors then that we could recognize our target consumer out of the.” (Jarvenpaa, Fiskars)
“I would of course really want really a lot want to see the data that kind of how our customers are uhm, moving inside the city or certain area.” (Koponen, Fiskars)
Areas of improvement were mainly, data accuracy in order to help avoid discrepancies in estimates and reality. Also being able to value right factors in the location selection process and give enough weight for specific factors so that mistakes in location selection can be avoided, in the specific example accessibility to the store was not valued high enough which resulted as a poor performing store.
“Centers that maybe the estimate s have been incorrect also for the traffic in population and traffic into that shopping center and that’s resulted of course then low traffic into our stores”
(Jarvenpaa, Fiskars)
“Location itself was good but then accessibility to the store itself was poor and then that wasn’t given enough value and then that was selected anyway and it resulted in not so well performing store.”(Jarvenpaa, Fiskars)
A clear sense that there is an understanding from both interviews that Big Data can be used in order to create a competitive advantage in the store location selection process. The participants believe that they can use Big Data to create competitive advantage by a) making better exact location decisions backed up by data and b) by saving both time and money by automating for example data collection specific to location selection process. IBM Institute for Business Value (2012) and Porter and Millar (1985) both give evidence in research that information through Big Data can be used to create competitive advantage for a firm. Thus this is supported and taken further that the competitive advantage can be created in location selection process supporting P6. Davenport and Harris (2007), support a better location choice through data optimization as a competitive advantage.
“Yes… Big Data as is as data doesn’t give the competitive advantage but if you know what you do with it and how you utilize it and how you draw the right kind of conclusions make right decisions based on that its for sure. For example in our case, where to put the store where to open a store, if the data shows that it’s the perfect place, then it most probably is the perfect place.” (Koponen, Fiskars)
“Yes, First of all if we didn’t have to manually do all the selection I mean that we could get the data from somewhere without having to count and estimate ourselves, instead we would get reliable data from some source that would help already.” (Jarvenpaa, Fiskars)
Overall this case shows differences in the answers widening the discussion about how much knowledge about Big Data use inside a company is at what specific level on the corporate hierarchy. There were clear similarities as well for example in what kind of direction they would like to go with Big Data and how this can create a competitive advantage for them in the location selection process.
5.3.2. Case Two
Carlson
Case two is a consumer goods company Carlson. Carlson was founded in 1859 now functioning as a subsidiary of Veljekset Halonen oy and currently these two have approx. 850 employees (Carlson 2017) and (Halonen 2017). Together around 30 stores are currently located in Finland. Two persons from the Carlson oy were interviewed for this case. The CEO and the Sales and Marketing Manager.
Research by Reno (2012), Manyika et al. (2011) and Gurstein (2011) can be linked to what is found in case 2. The company uses Big Data in their location selection process very little and they are not at the level in where they would want to be in Big Data use in location selection process. This is a well aligned answer in both interviews, as well as both mentioned that the Big Data that they do use, is utilized both out sourced partially as well as done internally. Proposition P1 is supported although the open source Big Data use is at the entry level. P2 is partially supported as the interviews showed that data is used both externally and internally really depending on the circumstances.
(Use of Big Data) “Partially but they are not currently related to our business that tightly, in
decision-making, very little I have to say.” (Heiskanen, Carlson)
“Uhm, very little. So not in the scale that we would like to. The data that we use is mainly historical data in its traditional form. We have on going projects, but we don’t use of what I understand Big Data or as relevant real time continually updated data. So only very little. We are in quite basic world in that sense still.” (Karhunen, Carlson)
“The ultimate analysis we do internally but most start from external sources. Either public or open sources, (tilastokeskus) or similar, also the statistics and such but we do also use bought data and information.”(Karhunen, Carlson)
“Yes of course we use, our own produced as well as gotten from other sources.” (Heiskanen,
Carlson)
The main reasons outlined why they are not using Big Data turned out to be different in this case as the CEO mentioned the reasons are cost and resources. Where as the other employee said they are not focused in this area now as the business is concentrating on other direction. This could be influenced by as was mentioned earlier that the data usage happens at the higher level of the hierarchy and that could be an explanation for the difference in the answers in this point. The answers do partially support P3.
“I think resources and cost of it. As main reasons” (Karhunen, Carlson)
“Well, really I can’t tell you the main reason, maybe well we are starting to invest into the online store heavily and moving that way” (Heiskanen, Carlson)
The interviews outlined that they use customer traffic and movement, populations, growth, and demographics, development values, average income levels, and business going in and out of an area in their location selection process. The most important factor for the company in the process is footfall and traffic as mentioned by both of the interviewees.
“When we talk about the store location selection process, we use customer traffic and movement, populations, growth and development values of areas, age, population division.
Uhm, demographics of the area, average income levels, business going in and out of the area and these kinds of.” (Karhunen, Carlson)
“Foot traffic, and parking is very essential. And in this business its very important that accessibility as a top factor, whether its trailer friendly etc. It doesn’t hurt also this kind of a ‘car dealership’ phenomena that other stores would be around and that’s natural for our business.” (Heiskanen, Carlson)
“If we look at over all then customer footfall and traffic. And also the purchasing power of the customer traffic.” (Karhunen, Carlson)
Importance of competitor location in case 2 is relatively unimportant as found in the interviews as seen in Figure 4. Contradicting theory and research in this area P4 not supported. This is debatable while the level is given an 8 in importance, compared to other factors it is not important. As well as this answer only is based on the CEO’s input. However also in this case customer data is given a higher importance over competitor data. Overall the ideas of the employees were well aligned the largest discrepancy is in the importance of foot traffic, and this is because M. Heiskanen is more in the hardware business were as the CEO provided a more overall image of their whole business.
