As part of our customer discovery process, we conducted in-depth interviews with more than a dozen enterprise CFOs about the innovation projects within their companies, particularly those involving artificial intelligence. Most are savvy about AI, have AI projects in play throughout their organizations (including multiple in the finance department), and categorize those projects into two distinct camps.
Their projects range from automating the supply chain to forecasting revenue more accurately, and while most are described as valuable, the two categories are vastly different, especially when it comes to return on investment. They are those that 1. automate manual process; and 2. optimize decision making. One CFO described the difference as the former “…does things faster than humans can” and the latter “…makes decisions humans cannot.” While the returns the CFOs described vary across all of the projects, there were two clear ROI camps: step-wise* and tangent-shifting*.
CFOs don’t just want a one-time uptick; they value systems that learn and improve over time based on employee interactions, global usage, and the data those systems process.
The projects in the step-wise camp are all about replacing manual process, and often involve investment in a general purpose robotic process automation (RPA) solution. An example is one of our pharmaceutical customers must obtain sales price, volume, and discount information from its distribution partners’ online portals. A bot automatically reaches out to each portal, gathers the information, and emails it to an email alias for people on an operations team, who then validate the information manually and enter it into the company’s ERP and financial systems. This process ensures that distributors’ sales comply with the company’s contracts and that the company can calculate its per-sale and cumulative net revenue at any given time. While it’s a time saver, it’s low value. With each project like this, the company’s return on investment shifts upward, but typically only once and in a limited way. Think of it as taking a single step up in value, but then moving forward horizontally over time.
Here’s an example of a project in the tangent-shifting camp: One CFO uses AI in retail to maximize probability of sale and price of merchandise, including whether to move inventory to a particular store, the sale rack, the outlet store, or out of circulation. It takes advantage of sales and other ERP data, demographic data, and exogenous data from online systems, and results in both revenue gains and savings. The difference between this and the step-wise project is that it generates value in a non-linear way, changing the actual value curve and shifting its tangent (its rate of change) not just over time but with each additional input (in this case, sales data, demographic data, and exogenous factors).
The CFOs who use the AppZen Platform to automate their spend auditing place it clearly in the tangent-shifting category, with immediate, significant, and rapidly-improving ROI as our AI models learn and the company benefits from connections to both their internal systems and thousands of external online sources, such as merchant and political databases, ratings websites, and social media. They describe using it to cut out a large swath of an onerous process, reduce erroneous, wasteful, or fraudulent spend before that spend occurs, and even change the culture and behavior of employees in a lasting way.
When we asked CFOs to describe what made these tangent-shifting projects unique and valuable, they surfaced the following salient attributes:
Business context. Citing the importance of a strong semantic layer in their AI solution, CFOs need solid business definitions and an organizational schema to put structure and consistency around the complexity machine learning introduces.
Local intelligence. CFOs know there’s untapped value in their internal systems like ERP. They look for valuable integrations and extensibility in AI solutions to use existing intelligence to make AI-assisted decision-making more valuable.
Global intelligence. In addition to data in their systems, CFOs are want to tap into relevant global systems ranging from websites to social media to online databases, such as those containing economic, scientific, and political information.
Meaningful changes. While conventional wisdom tell us that technology should wrap around existing process and not require big changes, CFOs attribute the most value to AI projects that allow them to make meaningful change to simplify process and change behavior for the better.
Constant improvement. CFOs don’t just want a one-time uptick; they value systems that learn and improve over time based on employee interactions, global usage, and the data those systems process. All the better if they can use the cumulative learning and system improvements to assist with the next use case, and the next, and so on.
In summary, our CFO interviews revealed a great deal about CFOs’ understanding, use, and value of AI projects. They’re savvy, thoughtful about how to evaluate an AI project, and have articulated the attributes of the most attractive AI projects, which emphasize time-to-value and long-term value.
*Note that “step-wise” and “tangent-shifting” are my terms, not theirs. During my conversations, we heard words like “small,” “incremental,” and “a double or triple” to describe the former, and “large,” “10x,” or “exponential” to describe the latter. I chose these terms deliberately after listening to many of the examples because I feel they’re most exacting to the range of outcomes the CFOs have described.