Talk to any modern business these days and you will most certainly recognise a repeating trend that’s hot in all industries and types of organisations: and that’s no other than data-driven decision making.
The volume of data businesses are accruing today is verging on overwhelming, which is why gaining a clearer understanding of how to apply big data analytics to business intelligence (BI) is now more important than ever.
The promise of getting this right and becoming a data-driven organisation is truly rewarding: higher ROI, reduced costs, optimized speed and processes, improved customer service and products, and many other benefits. Who wouldn’t be drawn to the analytics solutions?
However, the common concern that often slows down the adoption of data analytics is choosing the right type of solution for the business. Luckily, there are only four main types of big data BI that can really aid businesses in extracting meaning from raw data.
Descriptive analytics (what happened?) use data aggregation and data mining to provide insight into what happened, identify patterns and relationships that would not otherwise be visible. Data visualisation is often applied with this type of analytics to yield more insight.
Diagnostic analytics (why did it happen?) look at historical data to analyze past performance and determine why it happened. It is usually presented as an analytics dashboard.
Prescriptive analytics (how should we respond?) help answer the crucial question of “what should we do next?”, steering organisations in the right direction and highlighting the recommended next steps.
Predictive analytics (what could happen in the future?) use statistical models and forecasting techniques to deliver likely scenarios of what might happen in the future.
Descriptive analytics seek to capture Big Data in small nuggets of information in a format that would be easily understood by a wide variety of business readers. The main goal of descriptive analytics is to identify patterns and relationships that would be difficult to recognise within raw data and provide a historical review or summary of facts and figures in an accessible format.
Some of the most common applications of descriptive analytics can include summarising past events, such as seasonal or regional sales, customer attrition, marketing campaign performance, stock inventory, year on year change in sales and similar analyses. It can also be used to report general trends like most popular travel destinations, top-selling products and general market shifts.
Google Analytics now offer a feature called automated insights, which automatically surfaces important, actionable insights based on your data. For example, it could highlight a sudden jump in traffic and tell you where the new users came from. This saves you a lot of time clicking around different pages, trying to understand what’s going on and is all part of Google Analytics’ move towards “less data, more insights”.
Using data visualisations, in this case, the Vizlib Scatter Chart, can help you better understand the relationship between two quantitative measures. In this example, we’re looking into investment opportunities in GICS sectors, comparing Active Return vs. Active Risk.
Diagnostic analytics are all about identifying the reasons why something happened, allowing analysts to drill down deeper into the data and pinpoint dependencies and patterns. It is a crucial tool for companies that want to understand the factors influencing their KPIs, as it looks into and uncovers the hidden relationships and stories in the data.
For example, you would use diagnostic analytics to dig deeper into the data and understand why sales went up or down in a certain region where there was no planned promotional activity or which marketing efforts led to increased website traffic and more sales. Surfacing unusual events, detecting anomalies, identifying drivers of KPIs and recognising patterns are all key elements of the discovery phase enabled by the diagnostic analytics.
Here’s a fascinating example of the power of data visualisation. We have used the Vizlib Heatmap chart to drill down into the impact of introducing a vaccine against measles.
Prescriptive analytics literally “prescribe” the most suitable solutions to either eliminate future risks or take full advantage of available opportunities and trends. It is still a relatively new and rather complex type of analytics that uses sophisticated technologies like machine learning and algorithms and often relies on historical data as well as external information to extract meaning from big data.
The purpose of prescriptive analytics is to assess a number of possible outcomes and allow companies to choose the best possible course of action. A good example of prescriptive analytics would be predicting the next best offer for a customer based on their past purchases and the amount of money they spent with your company last month. You should use prescriptive analytics either when you want to explore and assess “multiple futures” or provide users with recommendations and advice on what action to take next.
Qlik’s Cognitive Engine is one of the coolest and most powerful tools out there that can streamline the prescriptive analytics creation experience. Just hot off the press, this new smart feature enables users to take raw data to deep insights in a matter of minutes (or even seconds sometimes!). By leveraging best practices and augmented intelligence to analyse your own raw data, Qlik Cognitive Engine can recommend data model associations as well as suggest the most suitable visualisation for your data. This means you can simply drag and drop your measures and dimensions to the design canvas or chart objects and the Cognitive Engine will automatically depict your data using the most suitable visualisation.
Predictive analytics use the insights of descriptive and diagnostic analytics to detect patterns, clusters, exceptions and tendencies and predict what might happen in the future. It provides companies with invaluable data insights that enable forecasting and predicting future trends. However, it’s important to remember that forecasting is just an estimate, not an exact science, and that it greatly depends on the quality of data and stability of situation.
Companies use predictive analytics every time they want to use the existing data to fill in the gaps in future scenarios and identify trends or possible outcomes. For example, predictive analytics are used to calculate the credit score that determines the probability of customers making future credit payments on time. When applied correctly, predictive analytics can be used to support sales, marketing and other business areas that rely on complex forecasting.
Qlik Advanced Analytics Integration enables users to connect any third-party forecasting tool to the Qlik environment. The direct server to server data exchange between Qlik Sense and the plugin allows for the data to be called from within Qlik script and chart expressions and calculated on the fly, making the visualisation of predictive analytics a breeze for everyone.
For instance, Qlik users can take full advantage of the Vizlib Line Chart to quickly and effortlessly visualise sales projections.
While no one questions the deep value of data analytics for any business, it’s important to note that organisations striving for actionable insights should focus on those types of data analytics that are most likely to offer the highest ROI. As Centrix Innovations explain in this example, predictive analytics are likely to provide the most business value but are also the most complex to implement.
Harnessing big data analytics can deliver huge value to businesses, adding more context to data ensuring it tells a more meaningful story. The type of data analytics that would provide the most value to your business will depend on the stage of your company’s development and the corresponding needs. In most cases, companies that strive to adopt a truly data-driven approach to decision making, typically combine descriptive and diagnostic analytics (a reactive approach) with predictive or prescriptive analytics that support a more proactive mode of big data analytics.