4. Discussion and Observations

Throughout this survey, we examine the trends and driving forces behind business-oriented visualisation. We identify a range of varying classifying features for the research (see Table 1 and Table 5). Ultimately, the goal for businesses is to generate a profit. This can be done through improving the efficiency of internal processes (internal intelligence), identifying the actions of competition (external intelligence and business ecosystem), or improving business-customer relationships and therefore increasing sales (customer feedback and customer behaviour).

The most popular data source was from pre-existing databases showing business ecosystem data. Rahul C. Basole is a significant contributor to this field, who is associated with over half of the papers in this collection.

The second most popular primary data source was used in the field of internal intelligence, demonstrating the relative affordability of internally generated data for the purpose of research. This is often qualitative study data that evaluate certain operations of the business. Internal operations are more easily accessible than external operations, and so produce more research.

Customer-centric visualisation literature has seen a shift from customer feedback to customer behaviour research (see Table 5). Prior to 2011, five customer feedback visualisation papers were published in contrast to just one customer behaviour visualisation paper. Post-2011, just one customer feedback visualisation paper was published and six customer behaviour visualisation papers were published. We speculate that this decrease in feedback-driven analysis and increase in behavioural driven analysis is strongly related to the increasing availability of GPS data from devices such as smartphones. This idea is reinforced through the timeline by which smartphone GPS data started to be utilised in research. The benefit of tracking customer behaviour over collecting customer feedback is that it provides an unbiased view of the consumer. Response bias can skew analysis and warps the decisions made based off of subjective data.

Our data classification reflects the business ethos of cost reduction. Primary data are expensive to collect, especially on a large scale. Even in the customer-centric fields where accurate, up-to-date information on the consumer is essential for successful business operations, we find that instead the researchers opt for web scraped data – sacrificing data quality for data quantity. This sacrifice in quality for quantity might be attributed to the advancements in big data utilisation. Instead of running costly studies and questionnaires that detail the thoughts and feelings of a smaller number of potential customers, a business can base their decisions on lower quality feedback from a large number of potential customers assuming that their data is representative of useful information and knowledge.

In the case of web scraping customer feedback, many data already exist publicly on the Internet. Creating a web scraper automates the collection and organisation process, substantially reducing the cost. While the quality may not be as high as primary data, the quantity often compensates for that. No secondary data sources are used in the customer feedback classification, presumably due to the time-critical nature of customer interactions. A company would not wait until the data already exists to analyse their target market. In addition, the nature of these data is very niche, i.e., it is likely that the data would only be of use to one company.

Webscraped data are seldom used for business intelligence. This is interesting as it highlights the relative distrust companies have for this data regarding the operations of their business. Primary data would be trusted the most, and secondary data sources can at least be validated. However, the unknown nature of online data is enough to prevent companies from deriving actionable intelligence from it unless the data are in the form of customer feedback.

Secondary data sources are far more popular across most fields. Particularly in business ecosystem research, which overwhelmingly uses pre-existing databases – presumably due to the ease by which existing databases can be accessed – but also due to the broad utility of business ecosystem datasets (see Table 5). The data are often used as a case study or proof of concept for visualisation techniques where the emphasis is placed on the visualisation techniques and not the business insight. However, it is still important to observe how the business data are being utilised through visualisation. If the visual design is successful in presenting the data in a meaningful way, then insight should naturally follow.

Table 1 highlights some gaps and trends in the data sources used in the field of business visualisation. The most notable trend shows that customer feedback data are heavily dependent on webscraped sources. The table also shows a significant proportion of the data used in the research is not publicly available. While this may seem disappointing at first, it suggests that the data were explicitly made available to academic researchers or the research was performed by professionals in industry. This increases our confidence that the visual designs were created with the goal of business insight in mind as the work was, at least in part, collaborative.

We planned a classification that identified the target user for each visualisation, however, the majority of the research did not specify who would benefit from interpreting the visual designs. Assumptions could be made, but not without some subjective guesswork. This is interesting because an ill-defined target audience suggests the research to be mostly experimental and in its early stages of maturity. This suggests that, if successful, the field of business visualisation could experience more growth in the future once the effectiveness of the field has been validated.

Other attempted classifications were also stricken with similar issues. We also looked at the different industries with which the data were associated. However, few trends were available using this taxonomy and often the industry was difficult to classify due to vague descriptions – possibly due to businesses wanting to keep details of their data undisclosed. We also attempted a "visualisation type" classification that based the structure of the survey on the visualisation techniques used in each paper. Though the taxonomy was not particularly useful, we observed that older research was more inclined to use 3D visualisations with more recent publications focusing on two-dimensional graphics – although the data are not conclusive. This topic was discussed at the IEEE VIS 2014 conference during the "2D vs 3D" panel. We already observe an increase in visualisation adoption (see Table 5 and Figure 28 on page 44) whereby newer publications are being published in non-visualisation journals, but still focus on visualisation.

This is a reliable indicator of adoption as business oriented journals are publishing visualisation centred papers.


Figure 28. This figure shows the growth of the Business Visualisation literature included in this survey.