Measuring ROI in AI Projects

By Rodrigo Rebollar, Specialist Lead at SDG Group.

One of the most common questions when approaching any type of project is whether it is profitable; that is, whether the expected benefit will compensate for the investment made and to what extent. This desire for measurement normally seeks to justify the development of an initiative within a company, but it also serves to optimize the allocation of resources, prioritize use cases, mitigate or adjust financial risks, or facilitate long-term decision-making.

As we can see, there are many benefits to having an approximate view of the return on investment (ROI) of a solution and, therefore, companies are increasingly demanding it be included as an integral part of the project itself. Usually, this ROI calculation remains a superficial estimate, since trying to cover it in its entirety is a complex exercise. 

In the case of technology and Artificial Intelligence, this return on investment measurement is even more difficult. It is common for many of the benefits to be intangible or difficult to materialize, for example, improvements in user satisfaction, agility in employee performance, increases in brand value, greater efficiency in strategic decision-making, etc. Not only this, but other factors such as difficulty in attributing the improvements to a specific initiative or project, limitations when estimating the real costs of processing or maintenance, and the lack of available information or the quality of the data itself, among other reasons.

To address these limitations, SDG has designed our own methodology that seeks to provide a broad approach, giving the most realistic understanding possible of the impact that will be had on the business. Based on our experience as a company, some of the elements that we incorporate and that we believe must be taken into account for the success of these processes are the following:

• Approximate the impact of the solution, not only from the affected process but from all interrelated areas. A single solution can have implications at many levels within an organization, and it is important to branch out all these effects. At SDG, we do this through what we call “ROI trees”.

• Define clear objectives that align with the business objectives and allow us to clearly understand which effects are considered successful and which are not.

• Consider (and design, if necessary) qualitative metrics that allow us to analyze the impact of the solution on the less quantitative aspects, and study its evolution before and after implementation.

• Compare the data obtained with industry or competitor standards to contextualize the return on investment and detect possible patterns or seasonality.

• Divide the implementation of AI into stages to be able to measure the ROI of AI over time.

There is a fundamental aspect that is worth highlighting and that is related to this last point. The calculation of ROI must be a living process, which evolves with the project itself and is adjusted as the information and data become more precise. The initial estimate when proposing the project cannot be the same as the one made once it is launched, or even after some time has passed since its implementation.

In summary, the ability to precisely and accurately estimate the impact and return of Artificial Intelligence and advanced analytics solutions is becoming an increasingly valued and sought-after asset for companies, as well as a differentiating factor of ours among customers.

Translated from Spanish. Original article on PMFarma.