![](https://www.raconteur.net/wp-content/uploads/2021/04/RAC_CFO2_IlloBanner-1.jpg)
Gartner predicts that AI software spending will surge to $298bn (£230bn) by 2027, up from $124bn (£96bn) in 2022. As investment in the technology balloons, finance leaders are under pressure to figure out if it will deliver on its promises of growth and productivity.
Most businesses assume that AI can boost their bottom lines, but figuring out by how much, and how long that will take, remains a challenge. In fact, research from software platform Orgvue found that 82% of firms are ramping up investment in AI, despite 50% being unclear on its business impact.
As the ones holding the purse strings, CFOs need to understand what AI can do for their business before they commit precious time and money implementing it.
Here, three CFOs share how they are assessing the cost, value and feasibility of AI projects in their organisations – and the challenges that come with that.
When it comes to implementing an AI solution, prioritisation is key. This is as much about deciding what you will do as it is about what you will not do. As CFO, you’ll need to fund and resource any new AI initiative, which means something else is not getting funded or existing expenditure needs to be reallocated.
Brokering the tradeoffs can be difficult but rewarding. It’s hard to say no to some ideas, but it will be necessary. This is where knowing your top priorities and goals is key.
In my experience, when assessing an AI project, CFOs and other C-suite leaders must first identify a goal or pain point and map all the steps of the process. Once this work has been done, and potential AI tools have been identified, it’s time to start evaluating the technical and resource requirements.
A crucial step in all of this is the partnership between the finance and tech department. CFOs should work with CIOs and their teams to ensure the new tools and processes are easy to deploy safely, quickly and responsibly, with minimal need for specialised skills or IT, data scientists and engineers getting involved. Complexity often leads to extra costs and slower adoption. Sometimes it can even make processes worse than before.
CFOs must also put thought into how to drive adoption throughout the organisation, knowing the nuances of each department and, most importantly, getting buy-in. Don’t underestimate the importance of change management in making the project a success.
At the end of the day, you will never have perfect information or a decision that is entirely without risk. Given there may not be as much certainty with AI investments, you will need to be willing to make an educated decision and an acceptable level of risk.
The first question on my mind as a CFO when it comes to a new AI project is: where am I driving efficiency? Is it improving a process internally or is it delivering value for customers? Once I identify the path or process, then I need to assess the various AI tools to work out how they can help plug the gap and generate a sustainable return on investment.
It’s essential to consider the specific business problem that AI aims to address and the expected results. Then you must measure the cost of the process using set metrics, for example, the average response time for a customer service call or the average time it takes to close a ticket in HR.
Crucially, one also needs to understand the opportunity cost of not embracing AI in a given use case.
At Commercetools, we’ve looked at tooling AI internally for a variety of processes. The biggest challenge that I have is the proliferation of these tools across different areas of the business versus having a consolidated view. It’s not about how many large language models you’re able to deploy, but rather how to get a few to do multiple jobs at once and be consistent with outcomes. Plus, you need to properly train and maintain the models as new data becomes available.
When considering implementing AI initiatives, the real challenge is assessing whether deploying AI will save costs in one area while potentially increasing them elsewhere. I’ve tried to solve these challenges by asking our leaders what they are currently doing, what they are going to continue doing and what they are going to stop. This is an efficient way of thinking about the feasibility and cost-versus-value approach to AI projects in business.
It is important to investigate the long-term role of AI within an organisation. As time goes on, AI systems accumulate a wealth of data and intellectual property, leading to important questions: is data structured in a way that allows AI to continue utilising it? How do I value this asset?
Based on the uncertainty we’ve seen in the economic climate in the past year, AI and automation is an area where many CFOs are concentrating their efforts. Although most finance professionals are known for being risk-averse, it’s no longer a matter of if they will be successful with it; they now should know they need it.
The most important question to ask when assessing the feasibility of any AI project is if your workforce is AI-ready. This means taking the time to evaluate whether your staff are equipped with the knowledge and tools to use AI technology in a safe and efficient way, one that complies with regulations and results in a positive return on investment. If not, this may require upskilling teams, which takes time and money, to ensure they understand the tools available to help them succeed.
It’s always important to think about the ultimate goal, as opposed to progress for the sake of progress. A worthwhile AI project is one that enables workers to better collect, process and analyse data, resulting in faster, more meaningful insights. These insights should, in turn, help people become more agile and improve their decision-making.
Some key areas we’re considering for AI project implementation, and likely other companies are too, include customer communications, such as sending invoice payment reminders, and financial evaluations, such as tracking KPIs around sales, revenue growth, cash flow and expenses.