5. T HE D ESIGN AND D EVELOPMENT OF THE R OADMAP
5.1 T HE PRELIMINARY PHASE
The preliminary phase contains the first five steps in which the initial conditions must be satisfied before actually starting the roadmap. The five steps need to be checked off before continuing to the subsequent phases. In the preliminary phase the support, sponsoring or capital, necessary conditions and boundaries of the roadmap are set. Without being able to complete these steps, the roadmap can and should not be performed (Garcia & Bray, 1997).
The first step is to obtain the support from management and other departments of the SME and to set the internal goals of where they want to go. These goals can be in terms of revenue as well as finding specific products for new markets in which the company fits strategically. Because business development is concerned with both the horizontal as well as the vertical coordination within the company, it is essential that all departments and people are addressed in the growth process to assure support throughout the entire process (Sørensen, 2012).
The second step is concerned with the different market scanning approaches based on the strategy of the firm (R. E. Miles et al., 1978; Olson et al., 2005; Slater et al., 2010). This roadmap is most appropriate for SMEs who hold an analyser strategy in which the company
acts as a fast follower of the prospector or product leader. Analysers monitor the reaction of customers on product launches of the product leaders and then follows quickly with their own similar or improved product. This strategy is low-cost while the prospectors take the initial risk of launching a new product or market while the analyser follows (R. E. Miles et al., 1978). Because this roadmap does not necessarily provide information about new technologies, customer intimacy or operational excellence, it is best applicable for SMEs which are looking for an expansion of their product portfolio without having to invest heavily into their R&D.
The third step is to satisfy the essential conditions. Although the external data used for the market scanning is freely available on Amazon.de, it is advised to scrape the data and place the information in rows for it to be analysed by the Tableau Software. Ideally, the estimation of sales is also carried out based on the BSR of the main category. There are two options to obtain the data. First, as mentioned before, there are several paid or free-to-use tools available which can scrape the web and store the obtained data accordingly (Mikut & Reischl, 2011). However, a certain degree of data scraping knowledge is required to automatically obtain the data on a weekly basis of thousands of Asins from selected categories do this (Schneider & Gupta, 2016). The second option is to obtain the data through a third-party IT-company. Scraping experts of professional IT-companies are able to provide reliable data due to their experience with the structure and possibilities within the Amazon.de webpage. Additionally, they are likely to be able to provide a SME with the estimation of sales based on the BSR and main category they want to scan. Lastly, the available data is best analysed with Tableau Desktop software. This software offers many useful functions and is, with the help of the market scanning manual, relatively easy to use. The steps that need to be taken for the data analysis are further explained in the market scanning phase.
The fourth step is related to the scope and boundary of the market scanning. When the previous steps are successfully taken then the scope and boundary of the market scanning are set. This step is essential to assure the quality of the results. Naturally, the Amazon.de is an external source which allows SMEs to scan in a broader manner (Haase & Franco, 2011; Liao et al., 2008). However, when the scanning is too broad, there is a higher chance of "paralysis by analysis" in which the obtained data becomes too much to handle (Kohli, 2017). On the other hand, when the scanning is not broad enough, important insights about consumer and market needs cannot be identified (Börjesson et al., 2006).
The structure of the categories and the different category levels provide a solution for this problem. Because of the hierarchal structure of the categories on Amazon.de, the higher the
level of the chosen category, the broader the information and product selection becomes. Therefore, a level-1 main category contains the top-100 best-selling products of the thousands of sub-categories and millions of products below while a level-6 category shows the top-100 best-selling products of a couple of hundred very similar products. Additionally, the main (level 1) category can be used for the estimation of sales from the BSR (e.g. (Cui et al., 2012) . It is therefore important for a SME to first select the main category or categories which fits best with the capabilities of the company and desired market entry. From there the sub-categories can be chosen based on the same criteria. It is advised to pick categories which are not below level 4 and include the above lying levels 1 to 3. This establishes a balance between a broad market scanning exercise while retaining specificity. Every category and sub-category is listed on Amazon.de itself and can be picked for analysis. It is advised to set the boundary and scope of the market scanning properly while this will avoid obtaining information from products that are not of interest. Furthermore, it is important to keep in mind that every extra chosen category adds approximately an extra hundred Asins to the analysis. Meaning, when 40 categories are chosen, the variables and information of around 4000 Asins are obtained and need to be analysed. The more Asins are collected, the more information about products becomes available but also the more time consuming the data analysis will be.
The fourth step should be taken thoroughly because it not only ensures a better time management in the subsequent market scanning phase but also establishes the first measure to find products that fit the company strategically.
The fifth step and last step of the preliminary phase can be taken when the data has been scraped from Amazon.de. Thereupon, the data needs to be cleaned from possible unwanted categories and products. Although the data can be obtained relatively easy and accurately (Schneider & Gupta, 2016), Amazon.de itself is still a dynamic platform in which products constantly appear and disappear in the top100 of a category. Additionally, products are sometimes added in a sub-category in which they do not belong and it is therefore necessary to clean the data first. First, the main categories (level 1) for which an estimation of sales has not been made need to be filtered out to assure that all analysed products are listed in the same main category. In Tableau this is done by "editing' the data source extract and selecting the
Top100_Main variable and filter this on the desired main category.
Next, it is important to set the criteria for the strategic fit for the company. In the subsequent phases there are several filtering steps wherein products that do not fit strategically in the company are filtered out. This is to assure that the quality of the results is highest and to avoid
the unnecessary analysis of non-fitting products. E.g. when a company is specialised in food supplements, it should filter out the products that have no relation to this industry or which the company is unable to produce themselves such as diapers, electronic toothbrushes. Although the "strategic fit"-criteria do not have to be set in stone, it is advised to have a direction in mind.
Subsequently, it is possible to already filter out products that do not fit strategically based on the prior set criteria. On Amazon.de, the higher the level of the category is, the more products are collected that do not fit the firm's portfolio or strategy. It is then advised to scan through these categories and collect all the Asins with a title that do not fit group these together as "Non- fitting Asin group " (see table 2: Tableau script "Data cleaning").
Database cleaning of non-fitting Asins
description The view is broken down by [Top100_Cat] vs. [Title] and [Asin]. The view is filtered on
[Top100 Cat], which keeps "selected main category of firm".
Marks: The mark type is Text
Rows Title, Asin, Non_Fitting_Asins_group
Columns Top100 Cat
Filters: Top100 Cat → Filtered on main category
TABLE 2:TABLEAU SCRIPT "DATA CLEANING"
By editing the Alias of the grouped together Asins to "non-fitting Asins" it is possible to filter these out of the dataset through 'editing' the data source extract. Select Asin Group 1 and exclude the non-fitting Asins member of this group. This initial step of data cleaning will make the visualisations more precise and in accordance with the SMEs portfolio and strategic fit.
When the five steps in the preliminary phase have been taken and the required data is obtained, it is possible to continue to the second phase, the market scanning phase. In this phase different steps will be explained in which the obtained data will be analysed.