Do all business intelligence solutions look basically the same to you?
It’s okay if your answer is yes. In a recent interview for my company’s podcast (Episode 24), Business Application Research Center (BARC) co-founder and CEO Carsten Bange explains why.
“The number-one reason why a company or person chooses a specific BI solution,” Bange says, “is actually now price and price performance ratio, and that is very new.”
According to Bange, this emerging price sensitivity likely reflects growing homogeneity across BI user interfaces and functionality. “The differences [between BI applications] have changed,” he explains. Now, the differentiating factors are “behind the scenes … aspects [such as] scalability, maintainability, [and] performance.”
“The number-one reason why a company or person chooses a specific BI solution is actually now price and price performance ratio, and that is very new.”
This means that those comparing BI solutions have to pay more attention to what’s under the hood and know what questions to ask. What follows is a primer on evaluating the backend of a BI solution that will help enterprises make better-informed BI purchasing decisions that hinge more on fit and less on price.
If a given vendor natively supports your database, this stage of the conversation is usually brief. If a direct connection isn’t an option, however, find out what connectors or drivers the vendor can offer to bridge the technological gap. Ask if licenses for these drivers are included with a subscription or come at an additional cost. Also keep in mind that drivers, which translate your data into a format usable by the BI solution, don’t always do a perfect job out of the box. Determine how many work hours your developers will need to refine the connection.
Note that some BI providers don’t connect to customer databases at all and instead require that all data be hosted on their servers. In such scenarios, it’s important to ask about the data transfer, or ETL, process and what it will require of your team. Additionally, and particularly if your company is required to comply with data privacy laws, take your time exploring security concerns.
Whether you’re a SaaS provider or a private enterprise, your data likely has some kind of tenanting structure. Your tenants may have either unique or shared databases, tables, and schemas, so find out what tenanting structure you have and determine whether that tenanting system is supported by the BI solution you’re evaluating.
When comparing BI solutions on scalability, be sure to ask about means, capacity, and associated costs as user traffic and demand for report and data storage increase. If hosting the application yourself, find out how easily and efficiently you may deploy an instance of it for new tenants. Check if it’s simple enough for a non-technical administrator to do or if the responsibility must be handled by a developer.
Performance is context-dependent and difficult to quantify objectively. Vendors may point to a chart depicting how quickly a particular report or dashboard renders under various degrees of strain, but this cannot account for the dozens of performance variables in a live production environment.
The best way to test application performance is by running it through a realistic scenario in your realistic staging environment. Build a dashboard your users would actually use, and use it to query the data they would actually query. Simulate a busy day by having a realistic number of users run the dashboard simultaneously.
Once you’ve performed these tests, it’s up to you and your team to compare BI solutions and determine if the results were satisfactory. Different organizations have different performance requirements. If a solution misses the mark, ask the vendor for troubleshooting help before striking it from the list because as there could be a simple fix.
Your security questions will depend on the scope of the BI vendor’s control of user sessions. If the vendor will be hosting your data, be sure to inquire after their database security protocols and compliance certifications. If the application is a standalone BI solution, ask about its user authentication and authorization protocols. If you have encrypted data, find out what steps will be involved in decrypting it for reporting. In all cases, find out what measures are in place to prevent nefarious acts and what process the vendor will follow in the event of a security breach.
Consider your current user and/or customer demographic as well as what that demographic might look like in three-to-five years. What languages, time zones, and currencies might need to be supported?
Consider also that a BI solution consists of four different content zones: static report text, dynamic report text (i.e., the report output/data), the database schema, and the user interface. All four need to be localized for a seamless experience, so probe for strengths and limitations in these areas when comparing BI solutions.
No matter how well a solution meets your needs today, it may fall short tomorrow when unforeseen use cases arise. To prepare, look for an extensible BI application your IT team could add to or modify using custom code. Find out what programmatic extension points the solution offers, what sorts of changes they facilitate, and how a developer would access them.
Finally, for embedded BI shoppers, there’s application integration to consider. As with performance, the best way to assess the power of an embedded BI solution’s API is to test it using a realistic scenario. If, for example, your administrators would like to use the API to set a database connection, load an authorization model, and set the user language, refer to the application’s API documentation to determine whether you can perform these tasks and, if so, how easily. (Keep in mind that tightly coupled integrations will be easier to implement but more difficult to maintain at scale, particularly as your application architecture becomes more distributed.)
Determine, also, how efficiently you may embed multiple elements onto a single host page. One of the advantages of embedded analytics is the ability to “pepper” a page with charts and tables. Check to ensure that each additional chart doesn’t eat away at your processing capacity.
The End Goal
Deploying an ill-fitting BI solution can be tremendously expensive, particularly if you’re locked into a long-term contract. Those enterprises that buy BI based primarily on price are much more likely to overlook a critical incompatibility and require a costly course-correction. Insight into the new BI differentiators will ensure enterprises against buyer’s remorse and help them find a solution capable of serving their needs for years to come.
Originally published with TDWI.