Overcoming Obstacles in Reaching ROI for AI Projects – with Fallon Gorman of NLP Logix

Matthew DeMello

Matthew is Senior Editor at Emerj, focused on enterprise AI use-cases and trends. He previously served as podcast producer with CrossBorder Solutions, a venture-back AI-enabled tax solutions firm. Prior, Matthew served three years at the World Policy Institute as a news editor and podcast producer.

Overcoming Obstacles 
in Reaching ROI
for AI projects@2x

AI initiatives that cannot find ROI have no use in modern business. However, the path to ROI from enterprise-wide digital transformations is never a straight and narrow road. A pre-pandemic survey from MIT Sloan Management Review and Boston Consulting Group found that 70% of companies among those surveyed reported no value from their AI investments. 

Business leaders must have patience and a thick skin when it comes to finding themselves without all the answers – which according to NLP Logix CFO and COO Fallon Gorman, is how they’ll be spending much of their time during the process. 

Gorman recently sat down with Emerj Senior Editor Matthew DeMello on the ‘AI in Business’ podcast to discuss the challenges and many misconceptions business leaders tend to have about the path of least resistance in achieving ROI from AI initiatives. 

Among the foremost challenges facing business leaders across industries today in driving digital transformations is judging how much it’s going to cost to implement, maintain, and continuously “care and feed” any automated solution, she tells Emerj: “Gaining a grasp on what it means to automate your business, as it relates to what it really costs you to do that process today is usually a pretty gray area for a lot of people.”

Fallon emphasizes that the process is hardly about micromanaging your employees’ ‘day-to-day’ but rather what insights can be derived from the aggregate data being taken from every relevant area of the business “where changes in process can provide lift, which is always a piece of it, and why it gets muddy on ROI. But also, where changes need to happen to enable appropriate automation to get to what the business is trying to accomplish.”

This article will examine three key insights from their conversation meant to help business leaders across industries find ROI from their AI initiatives: 

  • Connecting the symptoms SMEs feel with systemic pain points: Collecting and classifying sentiment data from subject matter experts (SMEs) on the severity of the inefficiencies and frustrations in their workflows to help business leaders understand the systemic roots of these pain points. 
  • Separating AI fact from hype: Addressing inaccurate assumptions about AI capabilities being popularized by the current media hype cycle by magnifying and clarifying use cases from areas of the organization with which leadership might not be as familiar. 
  • Not rushing solutions as the organization de-silos: Endorsing a leadership culture of patience, reflection and without rushing to conclusions, as the AI adoption process makes transformative changes to the business that will take time to settle. 

Listen to the full episode below:

Guest: Fallon Gorman, CFO and COO, NLP Logix

Expertise: Financial services, medical supplies, corporate management, accounting, financial reporting, SOX compliance, budgets, accounts receivable

Brief Recognition: Fallon began her career in finance at Bank of America while graduating from the University of North Florida in 2005. She served her first term as Chief Financial Officer at Medical Development International, LTD. Inc. shortly after. Fallon began as CFO at NLP Logix in 2019 and was promoted to her current dual role as Chief Operating Officer of the company as well in January 2022. 

Connecting the Symptoms Felt by SMEs with Systemic Pain Points

Often in digital transformations, companies focus on their pain points – or their most evident and problematic weaknesses in the business – but Fallon tells Emerj that such a focus often loses the forest for the trees. 

What’s required to find the authentic business signals behind pain points that drive ROI in digital transformations is to leave the rigorous and objective world of analytics and numbers and talk about what they are feeling subjectively: 

“It’s very touchy-feely [when we ask our clients] ‘How do you feel today about this pain point? What pain are you experiencing?'” she explains. “And teasing that out, I think to the extent that a business can understand their pain points, the easier it becomes to navigate the map of where you’re going to get your biggest ROI on anything.”

Companies often feel so many pain points that Fallon tells Emerj that the most effective mitigation strategy is to narrow the list down to the single most significant pain. From that perspective, not all subjective opinions of AI adoption team members are necessarily seen as being created equal.

Increasingly, the sentiments of SMEs and – to borrow a manufacturing term – “shop floor employees” on both business problems and proposed solutions tend to elevate above most other stakeholders outside of business leadership at this point of the AI adoption process. Finding the path of least resistance to ROI often means finding ways to record and classify their sentiments on problems and solutions in the data collection process.

Fallon tells Emerj that once organizations start effectively collecting these data from SMEs, the hard signals behind systemic roots of pain points – rather than singular incidents of inconvenience – start floating to the surface as those testimonials are brought to management’s attention. 

Fallon likens the difference to the same between a doctor noticing a patient has a cough and correctly diagnosing them with bronchitis. While SMEs are keenly aware of symptoms, Fallon emphasizes that only management will have the high-level perspective to diagnose the disease. 

Separating AI Fact from AI Hype

Almost needless to say, AI awareness among the general public is simply not what it was even nine months ago. Thanks to the explosive popularity of tools like ChatGPT and impressive text-to-image applications over the last year, every American teenager and kitchen table has some intimate, firsthand knowledge of how these tools work. 

While business leaders might be running out of excuses for not having familiarity with AI applications, the ubiquity of these technologies and the hype cycle currently running its course in the global media have led to an environment where it’s very easy to misunderstand the true capabilities of these tools. 

Fallon underscores that the best ally business leaders have in being able to discern real value from these technologies, again, is feedback from subject matter experts:

“Then you get to your subject matter experts who are talking about, ‘Yeah, so that sentence that ChatGPT just said would totally send us down three months of work that it’s not going to know that we would be going down that path.’ And that is where we find people getting caught up in the ROI conversation. A lot of it is – I hate to say it – trust but verify.”

That idiom is famously borrowed from Soviet foreign policy, but what Fallon means by it is that while SME testimony is critical – the subjective nature of their feedback needs rigorous investigation from management for its tangible substance. 

As an example, one of the more destructive biases that tools like ChatGPT engender in organizations, Fallon tells Emerj, is that leaders who don’t use the tools directly in their work assume it makes everything easier when that’s seldom ever the case, especially when the technology is first being used. 

Once management understands the practical day-to-day advantages of these technologies when used effectively, they may find more surprising outcomes on the path to ROI. In other words, just because you used AI and data to identify the problem doesn’t mean it will be the ultimate solution to the systemic issues at the heart of pain points across the organization.

Even more challenging for business leaders is how easily just the opposite can be the case, even with similar data signals coming from examining pain points:

“Sometimes, a lot of big ROI projects are not super complicated AI implementations. Clients and prospects think, ‘Oh because my pain and my problem is so big, this is going to be a really expensive problem to solve.’ Sometimes it’s not what’s causing that pain. It’s a really simple problem to solve from an AI standpoint, and sometimes the things that feel simple from a pain and problems standpoint are way more complicated, just because of those human decisions that are getting made along the way that have to be decoupled from the human to sort through how to make the right.”

– NLP Logix CFO and COO Fallon Gorman

Often trying to judge which AI capabilities will have such an inverse relationship between the ease of implementation, the complexity of execution, and vice versa tends to be a “chicken and the egg” game, Fallon continues. 

She cites intelligent document processing (IDP) as an example of a technology where it takes time to predict the ease or difficulty of implementation just from the appearance of straightforward use cases. Knowing that the smartphones in their pockets can quickly scan their faces, management might think an IDP use case reviewing contracts for essential information will be a seamless transition for their business. 

Conversely, management might have the false impression that a problem is “too big” for AI capabilities because they aren’t familiar enough with technology applications from other sides of the business not under their purview. 

“So in their minds, they’re thinking it’s this huge project,” she continues. “But there’s already technology that solved the problem; it just solved it in a different place, which is why they haven’t interacted with it.”

Part of these assumptions derive from how AI capabilities feel all-encompassing and, therefore, must require a significant investment in time and resources to manifest. To exemplify how misplaced these assumptions can be, the infographic below outlines the typical data infrastructure of an organization that has already completed its digital transformation. 

Note how small the hardware that houses the actual machine learning code is in comparison to the rest of the associated infrastructure:

Not Rushing Solutions as the Organization De-Silos: 

As the data collection and AI adoption process begins to gain awareness of pain points from all corners of the business, Fallon tells Emerj that management will gain a more holistic view of the business unmitigated by the silos of company divisions and the tunnel vision it engenders among employees.

She reflects on how communications channels have changed over the last twenty years – from paper mail to email and workplace messenger apps like Slack today. Where we’re likely to see similarly evolving AI capabilities converge into ROI for the business “is in the interdepartmental,” Fallon tells Emerj.  

In a previous episode in the same podcast series, NLP Logix Data Scientist and Modeling and Analytics Lead Ben Webster described an emerging data analytics discipline called topic search that looks at data signals no matter where the organization detects them. From a data-centric perspective, topic search is a way of looking at business problems and signals like Voice of the Customer, Voice of the Employee, or media surveillance.

Similar to how focus among business leaders is moving from Voice of the Customer to topic search, Fallon believes that looking at interdepartmental communications from a similar data-centric perspective will help business leaders better understand the systemic (rather than symptomatic) pain points at the heart of achieving ROI. 

As the process unfolds, Gorman encourages business leaders to let go of their knee-jerk tendency to reach solutions and ROI as soon as possible. She tells Emerj that often the fastest path to ROI is in deliberate reflection and acceptance of problems as they are and not forcing teams to come up with solutions that may waste time and resources to implement without guarantee of success:

“Oftentimes, people are their own biggest enemy in trying to get to that ROI. They’re trying to find a solution on the fly instead of working through the pain and understanding what the problems really are. So the advice I would give is: stop trying to come up with solutions all of the time, take inventory of the pain, and recognize that data doesn’t have to be something in a tabular database with which for it to be used. It just needs to be captured somewhere, preferably not sitting on something you have to scan – maybe move away from that order.” 

– NLP Logix CFO and COO Fallon Gorman

To expedite the process meaningfully, she continues, AI adoption teams should include experts in the process who have worked in AI initiatives in the past and have firsthand knowledge of how to achieve ROI in the specific use case at hand. Such experts should be able to contextualize the often contrary feedback from business leaders and SMEs and better realize a path forward based on that firsthand experience. 

Stay Ahead of the AI Curve

Discover the critical AI trends and applications that separate winners from losers in the future of business.

Sign up for the 'AI Advantage' newsletter: