As pressure increases for businesses of all sizes to leverage business intelligence (BI) for competitive advantage, there is a demand for new types of solutions to address the variety of business needs. As a result, a paradigm shift is emerging that opens up new approaches to the delivery and customization of BI solutions.


Historically, BI software products have been divided into two categories: highly specialized vertical-market software packages, and industry-independent general software that spans horizontally across virtually all industries.


These choices for delivering solutions - either horizontally or vertically - are now giving way to a new, third approach: diagonal solutions, which address business-specific functions that are relevant to several industries. Each of these approaches to BI meets the needs of specific markets or applications, so it is important to examine the role of each solution and determine where it might best fit.


Figure 1 illustrates examples of some of the different types of software applications available in the marketplace. In the retail column, for example, is size assortment planning, an application highly specific to retailers, and more specifically to clothing retailers. Similarly, in the “FinServ” (financial services) column, anti-money laundering applications are highly specialized.


Generally, these very specialized applications add a great deal of value to the businesses that deploy them and are priced accordingly. Typically, a service engagement is required to integrate these specialized software applications with other applications in an enterprise, which also adds to the price tag.



In contrast are horizontal market software packages, such as relational database management systems (RDBMSs), HR applications, online analytical processing (OLAP) tools, statistical analysis packages and data visualization tools. These packages are completely non-specific because they can be used by any industry. As a result, they also cost far less, and many of these applications are considered to be commodity software. However, if a company wants to use these software packages to solve a specific business problem, a thick layer of applications must be created on top of the base layer by the customer, or through a services engagement with a consulting firm or the software vendor. In this case, the actual price of delivering a solution significantly increases.


For example, a BI solution typically required the development of a data warehouse using an RDBMS, and on top of that, applications were developed using BI tools and SQL. This thick layer of application development adds substantially to the cost of any solution developed this way, and makes the deployment cycle long and the risk of project failure quite high. As Figure 2 indicates, approximately 80 percent of the cost of the implementation of a typical BI solution is spent on the labor required to create it.


Only 20 percent of the cost is in license fees for the underlying software. These solutions are built on horizontal-market software but end up being highly specialized due to the thick custom layer.



The software industry is responding to business needs to add high value through specialization, and also to lower the risk of extensive, customized application layers. However, basic economics require the business of creating and delivering these software solutions to be organized differently. Lowering the overall cost to the customer requires a larger marketplace than the tightly specialized applications typically address. The amount of customization must also be lower. Together, these require a shift from organizing software products only along either vertical or horizontal marketplace axes.


Consider instead software solutions specialized around common business functions. Quite different businesses often share common functions. If common business functional areas can be identified that have high added value, and also appear across enough industries for a larger marketplace, then those functional areas can be targeted for software development. They will be able to command a higher price point than commoditized horizontal market software yet still be sold in substantial numbers.


As an example, consider transportation logistics, which includes applications to help companies manage the use of trucking services. Many industries ship goods or equipment by truck, and optimizing the costs of shipping should be a first step in controlling business overhead. In the food segment of the consumer packaged goods industry, shipping costs can be as much as eight to nine percent of total annual revenues.


Figure 3 shows how transportation logistics cuts across the following industries: retail, industrial manufacturing and consumer packaged goods. These applications offer substantial vertical specialization and are applicable to some, but not all, industries. By contrast, FinServ doesn’t typically require trucking, with a few exceptions for shipping precious goods or currency.


The term diagonal describes the types of applications that cut across industries, though not universally, and that are also substantially specialized.



Businesses would use a solution in transportation logistics, for example, to accomplish several goals:


  • Ensure outgoing customer shipments are fully loaded.
  • Reduce intraplant shipments on critical shipping days.
  • Analyze costs, utilization and performance in different ways:
    • Year-over-year cost trends and
    • Comparison between carriers.
  • Identify opportunities to consolidate shipments to reduce carrier costs.

To address these goals, applications must be very specialized to trucking requirements. Some of the data concepts and metrics they must address include:

  • Line haul cost.
  • Cost/CWT (hundredweight).
  • Air weight (underutilized capacity).
  • Accessorial charges:
    • Fuel surcharge;
    • Detention - leaving a trailer at loading dock;
    • Layover - storing trailer for subsequent pickup; and
    • Unloading charges.

To optimize transport utilization, companies must have access to a variety of information and a granular level of detail such as accessorial charges added to the basic line-haul cost. Summarized data systems only enable users to identify excess expenditures; granular systems will allow them to drill down to the root cause to identify why there are excess expenditures so problems can be solved and costs reduced. Because accessorial charges aren’t visible on any summarized level of data above the individual bills of lading of the trucks, this system would require an integrated data warehouse containing the detailed information about every individual shipment.


Examples of diagonal BI applications also include sales, margin, and inventory analysis (see Figure 4). These solutions allow businesses to understand the costs they have based on the inventory they hold and their profit margins on products. A classic analysis that this type of solution allows is the ability to adjust inventory appropriately for high-margin products that are selling well. While any industry that involves inventory can benefit from this type of a BI solution, it is not a universally needed capability. FinServ, for example, shares the need for sales and margin analysis, but it doesn’t typically have inventory in the same sense as many other industries. In addition, there are other verticals, such as higher education, which do not need sales and margin analysis solutions at all.


The fact that these diagonal solutions are used across multiple industries makes the market for these solutions larger than a more highly specialized solution tailored to just one vertical market. This availability of a larger market is why diagonal BI applications are particularly interesting: the economics favor the creation of software solutions because they are both high value-add applications, and serve fairly large cross-industry markets.



There is a second, equally important and reinforcing phenomenon currently underway in the software industry that also strongly favors diagonal BI applications. This is the trend that Nicholas Carr identifies in his book “The Big Switch: Rewiring the World, from Edison to Google” (W.W. Norton & Co., 2008). This refers to the trend of businesses switching from the internal deployment of software to the outsourcing of large-scale information processing systems, also known as software as a service (SaaS).


SaaS is normally thought of as a simplification and economization of the installation, management and maintenance of software. Carr’s argument is simple: during the industrial revolution, every factory had its own power plant – initially, water wheels and subsequently steam engines. As electrical service became widely available, a big switch took place, and as we know, few factories today have their own power plants because it became more economical for them to use electrical power supplied by utility companies. Carr’s point is that in a similar fashion today, most industries operate their own computer systems but soon will switch to having them managed by outside parties because it is more economical.


Diagonal BI applications substantially extend the benefits of SaaS and will accelerate this outsourcing trend. Consider the kinds of large-scale BI solutions deployed today. As noted, these solutions typically involve a RDBMS and a thick layer of applications software developed on top of it using BI tools, data integration systems and so forth. The result of these projects, when successful, is a highly customized solution. Typically these are deployed in house, as illustrated in the top slice of Figure 5. The second slice of this picture shows how this same solution could be outsourced, that is, hosted by a third party that specializes in data warehouse hosting; however, the solution would remain very customized to the specific customer for whom it was designed and built. Hosting such a system externally just moves the burden of system management to an external provider, but it doesn’t change the amount of administrative and maintenance activity for the system. There is some economy of scale in this approach, but with diagonal BI applications, the economy of scale can be much greater because these functionally specialized applications share similar deployment requirements.



A SaaS-oriented software vendor has an opportunity to exploit this similarity by combining the information processing requirements for several customers onto a common infrastructure. For example, the data warehouse that underlies a transportation logistics solution can be identical in design and similar in expected workload across many customers who use that solution. In the bottom slice of Figure 5, I show that a single multitenant database can be applied to serve many customers. In a multitenant design, the data from multiple customers is combined into a single database instance. The needed identifiers are inserted so that each customer’s data can be queried separately from that of other customers, and the nature of the common diagonal application means that the workload behavior from all customers is similar. For example, database tuning that benefits one customer will benefit all customers.


Multitenancy is generally regarded as a key identifying characteristic for a successful and scalable SaaS-provider business model because it allows the cost of computing to increase far more slowly than the growth in the customer base. This is because multitenancy allows the computing resources on which the data warehouse is deployed to be fully utilized, creating an economy of scale that simply isn’t available when customized solutions are deployed on or off premises. Multitenancy also requires significantly less staff to service an infrastructure that supports a large number of customers. This economy comes not only from the SaaS deployment alone, but also from the synergy between SaaS deployment and the functional similarity created by targeting software solutions at market diagonal applications.


This same kind of economy of scale isn’t available via SaaS deployments for more general collections of BI applications, that is, nondiagonal ones. The workloads are so diverse that combining multiple customers onto a common multitenant computing infrastructure is far less compelling. For example, database tuning that benefits one customer’s performance will not necessarily help, and may hurt, performance for other customers.


This SaaS and diagonal BI economy of scale is of particular importance to smaller businesses. While large businesses might be able to achieve good utilization of their data warehouse systems and attendant staff internally, smaller and midsized businesses can’t typically achieve this on their own. However, with diagonal BI applications, smaller and midsized companies can benefit from the cost structure created from a SaaS multitenant deployment, which can achieve the same economies of scale that a large business can. As the marketplace of SaaS providers of diagonal solutions matures, these applications will be available to smaller businesses on a cost-competitive basis.


In conclusion, we can all expect to see a growing number of BI applications in the marketplace that match diagonal characteristics - specialized, but cross industry. Beyond this, the new SaaS software delivery trend will accelerate availability of this class of applications based on the synergistic nature of the functional specialization that diagonal applications offer combined with the multitenant SaaS deployment model.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access