Nothing exists in a vacuum. Any changes that take place will lead to consequences, whether the changes were planned or not. Business intelligence (BI) and analytics implementations are no exception to this rule.
Any project that gives access to data or analytics will have an impact on those who use it. The normal aim for BI is to help everyone across the business use analytics and improve their decisions. However, the act of providing analytics – particularly in self-service environments – should not be seen as the end goal in itself.
Instead, active use of data for decision making every day across the organization should be the overall objective. This creation of a truly ‘data-driven organization’ involves looking at more challenging problems that cover technology, process governance, compliance and business change issues. So how should IT teams plan ahead to get the most out of BI and analytics?
Starting the journey to full analytics implementations
Traditional BI projects undertaken in the 1980s, 1990s and 2000s would normally have been aimed at large enterprises that could afford the hardware, software and consulting time required for deployments. Today, the rise of cloud-based platforms means that many of these costs have been reduced or removed altogether. Cloud-based data warehouses such as Amazon Redshift or Snowflake can be used to store data for analysis, while new visualization tools have made it easier to import data for display.
However, in this journey to make visualization easier, one of the prime reasons why BI and analytics deployments were undertaken has been lost to some extent. In the past, BI was aimed at helping the CEO answer the ‘big questions’ that would have a fundamental impact on the business. More recently, the data governance that these big decisions required has fallen by the wayside in the rush to help more decisions get made, faster.
This shift should be viewed as a welcome one. After all, more people using data should equate to better decisions being made across the whole business. However, new pain points often erupt when people use data that does not match.
For example, a sales report using figures from Monday afternoon might be very different from a report using data extracted on Thursday morning. Similarly, the head of sales may think of revenue based on orders signed, while the CFO considers revenue strictly by invoices sent. These different interpretations of the same term – ‘revenue’ – can lead to different analytics research being carried out, and ultimately incorrect decisions being reached.
This kind of error makes it more difficult for analytics to deliver value for the business and keeps it from getting adopted more widely. Paradoxically, the spread of data has not been matched by an equal spread of insight, trust and improvement.
BI and analytics programs should be designed to provide everyone across the company with access to the same data in a controlled and managed environment. Taking this approach should enable everyone to answer their own ‘big questions’, rather than keeping analytics as the sole preserve of the management team.
Why new approaches to BI should be about pulling data, not pushing reports
The long-term goal for BI and analytics projects should be to get data into the hands of everyone who needs it across the business. However, this goal is not simply about extending reporting and data discovery to more people across the business. This simple aim covers a lot of business change management issues that should be considered alongside the deployment of new visualization or analytics tools. Areas such as supply chain management, sales and marketing can all benefit from networked data. Indeed, being able to collaborate around data sources can open up new opportunities to improve performance.
For example, supply chain teams don’t often collaborate with marketing, yet both are involved in delivering a great user experience and getting the company’s product into the hands of customers. Knowing the real costs involved in delivering additional products to stores, alongside a demand campaign, can ensure that these initiatives are more profitable over time — rather than leading to additional costs that wipe out those gains. By collaborating and sharing data, both teams can profit.
Building up BI and analytics usage within the company has to start somewhere. For companies with some existing data analytics capabilities, this can involve looking at the current BI implementation and establishing if it is fit for purpose to be rolled out to other employees within the organization.
To evaluate this involves looking at what is in place and working. Typically, companies will have some central reporting and analysis for corporate and financial governance, while departments may have bought in their own data discovery tools as well. This mix of central and local tools can meet specific needs, but not the whole approach to using data. Instead, it’s important to look at how to blend the central approach to data governance with the flexibility and user friendliness of local tools.
From personal experience, once users get access to analytics and can answer their own questions – rather than having answers provided by a third party – the demand for data access only goes up. For IT, this increase in the use of analytics within the business represents an opportunity and a risk: on one hand, there is the chance to help departments collaborate and improve their performance based on the use of data for decision making; on the other, each department may try to go their own way. At the same time, implementing new central services that require additional reporting and analytics staff does not represent an efficient use of budget when users are more comfortable with self-service options.
This puts more emphasis on designing a scalable approach to managing and governing data so that requests for analytics can be met without additional IT overhead. As more users require access to data, the need for self-service will grow to avoid this management headache, too. Rather than pushing data to users, it’s worth examining how users can pull their own data together while keeping governance in place.
This approach depends on central data from systems of record, such as finance and operational ERP, being available alongside local or individual sources of information. These different sources can then be networked together to build more specific reports or dashboards. For example, the marketing team might need information on sales performance and invoices paid alongside their own campaign data. By using this data together, marketers can see how successful each campaign was, in terms of actual revenue generated.
Networking these different sources of data means that each team – and each person within that team – can then use this data to meet their own needs. For example, while a Chief Marketing Officer might care about lead-to-cash and marketing campaign impact, a channel marketer can go into his/her data in more detail to see which campaigns generated the most revenue and where others underperformed. The same data can be used to provide different analytics results and dashboards, but the information underneath will all be consistent.
Sharing analytics results
One big difference that can come from this kind of exercise is around how data gets shared within the business. Traditionally, this has been done using static reports in PDFs or with spreadsheets. While spreadsheets can be used for some analysis, the only real changes that can be made are what the spreadsheet software allows. This is better than nothing at all, but it is hardly a way to produce great insights.
Bringing BI and data sources closer together can also help share the results of analytics in new ways. Now, the whole analytic environment can be shared with another person. Rather than just sending over the results of the analytics, the dashboard and data can be shared for users to collaborate on. This approach can provide them with insight into the thinking behind how the analysis was put together and the data sources used, not just the result.
This approach to networking data sources together can also enable companies to deploy ‘analytic incubators’. These are ‘sandbox’ environments that support autonomous experimentation and rapid prototyping by end users, while greatly minimizing cost and risk. These incubators can become an effective way to uncover value produced by users throughout the organization and then share insights with others.
For companies that want to innovate around analytics, incubators can provide good opportunities to build out awareness of analytics across the company. Putting incubators in place can also provide opportunities for collaboration between business groups that don’t normally talk to each other.
At the same time, this approach can minimize the risk involved with users misapplying analytic techniques. It also prevents another burden being added to the central reporting team. If this self-service approach can’t be adopted, then more staff may be needed to keep up with demand.
Looking ahead, the challenge for BI and analytics implementations remains the same – how to help business leaders make better decisions through using data. However, the deployment of cloud-based platforms has created an opportunity to make more data available and useful to more people across the business. While the vision remains the same, the number of people that it can apply to has gone up massively. BI and analytics technologies have to deliver on their promises, and that involves making analytics work in practical ways for everyone. Networking data together is the required first step on this path.
The preceding article was originally published in IT Pro Portal on September 1, 2016.