Approachable analytics: one size doesn’t fit all

With the current digitizing of all industries, data gathering and data analyses are very much top-of-mind. Consequently, having data analytics within easy reach of everyone is also gaining in importance. There are 3 good reasons why approachable analytics is back on the map: organizations want more insights, want to shorten the time to value and embrace the self-service approach.

In the current market conditions, organizations have to be able to react fast. Making correct and sufficient real-time decisions is only possible when having good insights within the organization. That requires more than just having a clear look on past performances, says Hylke Visser, Principal Business Solution Manager at SAS Southwest Europe: “Good decisions aren’t just based on what we already know, but also take future predictions into consideration. There, we notice a clear difference between traditional BI tools – only looking into the rearview mirror – and innovative tools that include analytical and statistical capabilities for solid forecasting. More and more companies are putting their resources into those analytical tools.”

In order to be more agile, organizations also desire more agile tools. Scrum and other agile methodologies are quite a hype today. That’s because they address the shortcomings of traditional systems such as long implementation cycles. In an age where search engines can answer almost immediately questions about information on the internet, business users are longing for the same user experience for their corporate data. “We want to be able to start working on an answer when the question arises, and not first having to create a Request for Change”, said Visser.

The willingness to adapt approachable analytics is also related to the costs associated with traditional BI projects. “Letting people explore data in a self-service way reduces the IT costs and, at the same time, it fulfils the expectation of having more insights and a shortened time to value. Especially around insights, we are seeing that business users have a good feeling for what they want to explore and discover in their data.

An integrated, in-memory, analytical platform

An integrated, in-memory platform that combines the traditional BI world with predictive and descriptive capabilities – such as SAS® Visual AnalyticsTM and SAS® Visual StatisticsTM – addresses the needs of all system users. They allow of more autonomy and a solid data governance across the different layers.

The bottom layer of such single platforms is the traditional BI layer. “Many users, both in and outside the organization, can digest information about what has happened in the past. Typical examples are dashboards with KPIs and charts that can be consulted via the cloud and mobile devices, but interactively too in the Microsoft Office productivity tools such as Excel and PowerPoint”, said Visser.

In the second layer, people can extract the information that matters most to them and tailor the charts and dashboards to their needs. This self-service layer speeds up the response time from business users to their internal or external customers.

The third layer – the `data discovery layer’ – allows business analysts to find new correlations in the dataset of the organization. Exploration techniques help to find the pea under the mattress. Some really good methods there, are Sankey Diagrams or decision trees.

The real data scientists will find their likings in the top layer which enables analytical modelling using techniques such as clustering. New identified clusters can be reused, in turn, in the traditional reporting environment.

How to start with approachable analytics?

Organizations don’t start with approachable analytics from a BI replacement perspective. Approachable analytics demands a mental and cultural shift within companies and, therefore, it is best to start the activity within an innovative area or division of the organization. Making some quick wins before scaling the project to the entire organization, supports the business case. On top of that, a division that’s focused on innovation is more open and receptive to out-of-the-box thinking. “We see that our single, integrated, analytical platforms are often used to enrich the existing dataset with new, open source data, such as the weather forecast or social media conversations. This in order to apply the data more quickly to the market demand, improve the performance or develop new business models. For example, a collection agency uses our single, in-memory solution to add external information – like from the Land Registry – to their existing dataset. By making a quite solid estimation of the funds and resources of the people with outstanding debts and combining that with analyses of historical behaviour, they define their working method. In some cases a phone call can be the most efficient way to motivate people to pay their outstanding bills, in other cases sending a bailiff is the only solution”, Visser explained.

Also important is to embed approachable analytics within the organization. The appointment of a Chief Data Officer can help a lot, as would creating an innovation department. Data-driven organizations that have made this steps, are typically more successful. By thinking more about the value of analytics from the very beginning of the process, data can be used in a more efficient manner.

Besides that, a lot of people underestimate the benefits of visualization and data discovery. “Excel-sheets feel very confident and capture data well, but data visualization truly shows the power of data. In that sense, a data visualization workshop of a few hours can be a real eye-opener,” Visser concluded.