When chief information officers (CIOs) are proposing new IT initiatives to the board, one of the first questions asked of them will inevitably be what return on investment (ROI) the company can expect to see from the project.

However, this is a calculation that is often not easy when it comes to big data analytics, and in many cases, CIOs will have to admit they simply do not know how to answer this question.

InformationWeek writer Paul Korzeniowski stated ROI is the most common justification for technology projects, with executives wanting to know about expectations for payback before they sign off on any significant spending. But the potential benefits of big data are often uncertain at the outset, so it can be impossible to offer any certainty on this.

Developing big data applications is often an iterative process that requires patience and a degree of trial and error before businesses find success. Nick Heudecker, research director at Gartner, noted: "A corporation may start down 19 different tracks before hitting pay dirt on the 20th attempt."

As a result of this, CIOs need to be thinking beyond ROI when it comes to justifying their big data projects, while also ensuring they are not jumping into initiatives blindly.

IT departments should look to work with business units and persuade them to start with small trials that can provide a proof of concept, before extending them to wider-scale solutions. 

Samer Forzley, managing director at Marketdrum.com, observed that in many cases, organisations should be able to draw on their existing pools of data in order to limit initial investments, rather than going out and building an entirely new infrastructure straight away.

Once big data projects are up and running, however, the type of insights they can deliver are typically able to lead to much greater efficiencies across many areas of a business – often in areas that traditional ROI calculations have not accounted for.

For instance, big data can be highly useful in helping organisations evaluate their processes and inform them whether any steps in place are really there because they are the most efficient way of doing things, or just because managers have made assumptions about what constitutes best practice.

Mr Korzeniowski said, for example, that an energy company may assume the best way to keep its transformers working correctly is to send an engineer to perform routine maintenance every six months.

"With Internet of Things technology maturing, the utility firm can now collect granular usage and performance information on the devices and develop big data applications to determine if those assumptions are correct," he continued. If these solutions can identify a better way of doing things, the potential benefits can be significant, but there is no guarantee at the start of a big data deployment whether this will be the case.

"Big data projects require that CIOs think outside the box and rewrite the traditional ROI rules they have followed when deploying new applications," Mr Korzeniowski said. "The payback comes near the end, rather than at the start of the project."