Making decisions with data: How to put different data sources together

There is a huge amount of hype out there at the moment around analytics. However, there is good reason for all the excitement and investment that is taking place.

The wide availability of cost-effective cloud computing services makes it easier than ever to store data and to work with it in new ways. This, in turn, means that many companies are now looking more closely at the data sources they hold and how they can use this data to improve their operations.

In its FutureScape for Big Data and Analytics, IDC predicted that growth in applications incorporating advanced and predictive analytics, including machine learning, would accelerate in 2015. Applications that make use of analytics will grow 65 per cent faster than apps without this kind of functionality, according to the firm.

The reason is that companies can make more use of the data held in applications today – both through providing new insight into existing data within a given application, or more interestingly by combining it with other internal or external data sets.

At the heart of this predicted growth is the assumption that the availability of more data can help people make better decisions. Now, decisions made with data should be more successful than those made on hunches. However, putting that theory into practice is the challenge. For IT, BI and analytics teams, it’s important to understand how people across the business will actually understand how to use data in the right ways to support their requirements on a day-to-day basis.

Preparing for the move to data-driven

There are several areas where companies will have to rethink their approaches to analytics and use of data, and some of them will potentially overlap. The first is to understand how using data will impact job roles in the future. This requires a fuller understanding of how people across the business currently make use of data, from personal decisions through to wider planning and decision-making.

The aim here is to get a benchmark for how the company – and the individuals within their operations departments or business units – uses data. Without this baseline, it’s more difficult to see how data usage improves those operations over time. Equally, the team involved in any current analytics or BI implementation can show what elements have been successful and then look at how to tie their projects into wider data strategies.

Alongside this level-setting, developing a central strategy requires input from those decentralised business units and departments on how their operations can be improved. This is important, as there are huge amounts of knowledge and good ideas within every business about how to exploit data, if employees are given access to information sources in the right ways. It also stops the central team from mandating how data is used by each functional team, or by all the individuals within a business unit.

Central edicts on how to work with data can fail for two reasons. The first is poor change management, when employees don’t understand why there are changes being made to how they work. Without strong change management, this can lead to backsliding into old habits . In the world of BI, this can mean that decisions will still get made on hunches or incomplete data, rather than using the new services provided. When implementing a new analytics product or embedding data into an existing decision process, it’s important to bear this element of change in mind, so that people come with you on the journey.

The second is more painful – the central team implementing new business intelligence or analytics solutions won’t know everything about the business unit that is involved. Let’s take an example around sales operations for an international company. While the business may have to report revenue across its operations in one standardised way, there will normally also be differences in how each country unit or regional business may roll up its figures to meet its own reporting and compliance regulations. While this might seem a minor point, it can lead to frustrating discussions around correct numbers for sales volumes and profit margins when people aren’t working with the same data in mind.

To overcome this, it’s important that everyone can work and understand the data that they have in their own ways. Central operations teams can and should own the wider data sets that support global reporting on sales, profit and loss. However, that data should also support the local team that has to report its own numbers. The difference here is that the local team can and should be able to look at its own data in context, so that the right decisions can be made. This flexibility is key, as a “one size fits all” approach to data can miss some of the nuances that are required.

Central and local data

Preparing the ground for using data across the business can be a big exercise in itself. Many companies will have existing business intelligence or data programmes, while others will want to integrate new sources of information into their processes. For all companies that want to become “data-driven” in their business operations, the ability to bring this data together is essential.

What this means is bringing together all the different sources of data and delivering a unified view to users. This can involve physical movement of data or logical connections that allow data to remain where it resides but more importantly users feel like they have everything they need for their analytics in one place. This makes it simpler for analytics to be carried out across data and potential correlations created. A good example to look at is how the role of marketing is changing based on greater access to data.

For instance, a customer may have records on them in multiple systems, ranging from the Customer Relationship Management (CRM) system and sales records, through to the finance, Enterprise Resource Planning (ERP), customer support and service desk applications. Each of these separate systems will have a slice of data on the customer. By linking up these sources of data, it’s possible to get a better view of the overall customer lifecycle.

This is very important for managing marketing and sales programmes. To gauge the effect of digital, social and web marketing campaigns, marketers might have huge amounts of data on what customers like or download. However, it is only by integrating with sales and finance that marketers can link their activities back to revenue achieved. What looked like a successful campaign with great engagement might not lead to any uptick in sales. Without the right data in place, it is difficult to know for sure.

As marketers get to grips with data from across the business, there is often a realisation that their current metrics are not suitable. Rather than looking at engagement numbers, tracking the success, in terms of bookings and revenues, is much more relevant and useful to the business. Metrics such as customer acquisition cost, customer lifetime value and lead-to-cash can be created and monitored.

Now, for companies that may have multiple offices or country teams, not every department will report things in the same way. There will be local variations and requirements to bear in mind. It is possible to host and control the necessary data centrally, but also let this be augmented with external data from local sources, whether held in a simple spreadsheet or provided by a partner.

What is important is that any good ideas that are developed around use of data are not just held by that local team. Those ideas can actually be shared and re-used by others within the same organisation, as the activity is all linked back to the central data.

From a data perspective, this ability to “manage global, act local” around data can be a huge source of advantage. By providing access to data to everyone, the scope is there to improve decisions across the whole business.

Pedro Arellano, Birst.