Business intelligence: managing data complexity with analytics
Exploring the “why”
With Sisense now a part of its solution, Profit Tools can offer more insights to its customers. A typical scenario would be for a customer to receive corporate profitability data that differs from the level of the previous month. “The analytics layer provided by Sisense allows our customers to find out what component costs are affecting profitability,” Widell explains. “They might find that the fuel cost is the same, but that unbilled waiting time has gone up. Then they could find out whether the drivers are being delayed at a port or rail terminal or at the customer’s facility. Once they know the cause, they can start considering options for solving the problem.”
Looking at profitability or enterprisewide KPIs would be a typical starting point for an executive. “Any good dashboard raises questions,” Widell says. “The power of an analytics solution is that it can explore why something is happening.” At the regional level, a manager would be looking at a more limited issue than corporate profitability—for example, whether in general, shipments were arriving on time and in good condition. At an operational level, the appearance of out-of-range service metrics would mandate a near real-time answer for the root cause. “Ten years ago, it might be difficult to even detect an issue, much less get to the root cause in real time,” Widell adds. “With the analytics from Sisense, our customers can achieve both goals.”
The Sisense platform is well suited to complex data, according to Guy Levy-Yurista, VP of strategic growth at Sisense. “The volume, variety, velocity and volatility of today’s data, especially data originating from IoT sources, led us to focus on solutions for complex data,” he says. “The Sisense platform has an open single stack that can handle data end to end, from connecting and cleansing to collaboration, and it achieves a very high speed for analytics by using cache memory rather than RAM.”
Data does not need to be collected in a data warehouse but can be analyzed from its existing databases. A semantic layer reconciles issues such as different field names for the same type of data. “We leverage existing structures—users just need to connect to the sources,” explains Levy-Yurista. That capability is particularly helpful for cases such as intermodal shipping, which has data scattered throughout many repositories, requiring near-real time analysis to allow for adjustments in operations.