Let’s say your company has decided it’s time to take the leap into the 21st century and implement business intelligence software to start making smarter use of its data. While most executives who believe in a data-driven approach would agree this is a good idea, many business intelligence initiatives either fail outright or fail to produce the desired, easily identifiable ROI that organizations expect to see.
One of the main reasons that BI solutions fail to live up to expectations? The buyer does not fully understand the true costs of business intelligence. As Gartner notes, “BI leaders often focus solely on license metrics when comparing BI platform vendor and analytic tool costs, even though it makes up only a small portion of overall cost of ownership.” Thus, they fail to properly evaluate the pricing and ROI of the tools they implement versus alternative offerings—either a competing tool or continuing to operate without a centralized analytics platform.
The problem begins when buyers look at the “sticker price” of a BI solution and compare it with the direct returns they might receive from data analysis. However, this type of thought process can lead to severe misconceptions which can run in both directions: either underestimating the costs of the current situation, or those of the BI tool that is being considered.
Let’s look at three of the most common hidden costs of business intelligence, and then suggest one surefire way to avoid most of them.
1. Scaling Pains
Let’s say you’re currently working with two data sources (e.g. CRM software and web analytics) and generating reports on a weekly basis using a spreadsheet tool such as Excel or Google Sheets. While this process is bearable, it is time-consuming and error-prone, so management brings in a dashboard tool to help automate these reports.
In these situations, the temptation is often to purchase a low-end tool that can answer the pain at hand. This makes sense initially, but can be problematic when it comes to data—because like it or not, your data is very likely to grow in a year or two. As the world becomes more digital and more measurable, the number of systems that create relevant data increases. So does the amount of data.
Your one million rows could easily become ten million. Furthermore, new digital systems and IoT devices could lead to significantly more data sources and complexity. This can be seen in a recent study conducted by Cisco, which estimates that by 2020, the total amount of data created by IoT devices will reach 600 Zettabytes per year. For reference, all the information sent over broadcast in 2007 was about two Zettabytes.
At this stage, the bargain-bin tool that you previously picked up to solve your Excel woes turns out to be a crutch: it can no longer answer your data needs (in terms of performance or complexity of analysis), but on the other hand it is already entrenched in your organization. Licenses have been purchased, training hours spent, and employees are hesitant to learn how to use yet another software tool. Thus you find yourself stuck with the unsavory dilemma of spending additional resources to implement a full-fledged BI tool or sticking with a solution that does only half of what you need.
2. Integration Costs
An important consideration in any cost-benefit analysis of business intelligence is how the software you’re looking at will fit in with your existing ecosystem, and whether it will provide a standalone data analytics solution or need to be part of a patchwork of additional programs in order to produce real value for your organization.
The important thing to understand here is that there are several parts in the analytical value chain: the first is connecting to the raw data sources (databases, files or cloud sources); the second is the ability to perform ETL (extract-transform-load), cleanse and model the data in preparation for analysis; visual analytics and ad-hoc queries are the final stage that depends on the success of the previous ones.
However, many of the BI tools in the market focus on that last stage—with ever-flashier graphics and dashboards masking the fact that all the back-end “heavy lifting” has been delegated to a separate tool or to IT professionals. Hence it is important to understand exactly what functionality you will be receiving from any type of software you’re considering. Full and single-stack BI tools are meant to be used as a platform to handle everything from data preparation and modeling to creating and sharing dashboards; other tools focus on richer dashboard functionality while giving very little benefit in terms of preparing the data.
There’s nothing inherently wrong with using data visualization software along with proprietary ETL and database tools—but it is important to understand that this will affect the overall price you can expect to pay to achieve the analytic solution you’re looking for.
Lastly, you need to consider the price your organization is paying when valuable employees are spending time generating reports, rather than attending to other mission-critical tasks. Obviously, this is true in the initial scenario, where everything is being handled through spreadsheets; but even when procuring a modern BI tool, you still need to make sure it can adhere to the standards of self-service for business users.
Traditional enterprise business intelligence was built with an IT professional or experienced data analyst in mind—one who would feel comfortable using scripting and coding to create a new query. Modern tools try to avoid this, but often their lack of back-end functionality will require this scripting and coding to take place in the preliminary data preparation stage.
The result: instead of business users struggling with endless spreadsheets, the burden has been shifted to the technical workers, who now need to operate IT-centric systems to generate the same reports. To prevent this, you want to see that the business users in your organization can actually use the tool that is being considered and that they are able to answer their own data questions without constantly required internal or external tech support.
To Reduce Your Risk, Ask for a POC
While the above might seem discouraging, there is a simple solution to avoid the “gotchas” and hidden costs associated with some business intelligence solutions: proof of concept. If a vendor claims their tool is easy to use, fast, agile, or anything else—ask them to prove it, on your data and in your environment.
This is true for most IT systems, but particularly for sophisticated software such as BI tools. Insist on seeing the system actually produce at least some of the desired results before opening your wallet. This will give you a clear idea of the system’s ability to handle your existing needs and scale; you can see how well the tool can integrate with your existing systems and data sources; and the people who will actually be using the program can start playing with it to see how simple it actually is. Understand these factors, and you’ll make a much more informed purchasing decision that better reflects the costs and savings of your BI solution.
The flipside of this is when the vendor refuses to run this type of proof of concept or asks for an implementation fee. While this might be justified in some extreme scenarios, it is definitely not a great sign. Whenever possible, opt for a proof of concept on your data rather than a demo on sample data. This insistence will benefit you in the long run and save you from the pitfalls of a lukewarm BI initiative.
Are you looking to improve your employees’ performance with business intelligence software? Keep these hidden costs in mind while shopping, and you’ll be better prepared to buy a solid product. Check out Capterra’s business intelligence software directory to start your search.