AI Initiatives are Investments. Act like it.
The importance of understanding Capital Cost as an AI Product Manager
#beyondAI
While an AI Product Manager won't promise to make the company rich 💰, they do guarantee to minimize their losses! 💸 - I often tell this to new AI Product Managers and companies thinking about adding an AI Product Management role to their team.
Behind this statement lies a crucial idea: every AI initiative is, ultimately, an investment, and failing to approach it as such is the first mistake one can make.
Neglecting to view your AI initiatives as investments can lead to selecting AI use cases that fail to yield a return. This oversight can result in not just financial loss but also damage the trust you've built with your company and customer.
To better understand how companies assess investment opportunities, in this issue, I will discuss an important financial concept and its significant role in the realm of AI Product Management.
You will learn about:
The average ROI of AI initiatives compared to top performers.
What Capital Cost is and why it's crucial for AI Product Managers.
Case Study:
How Capital Cost as a benchmark is used to evaluate AI initiatives.
AIPM Thoughts on Case Study
Final Thoughts
Takeaway for Companies
Takeaway for AIPMs
Have fun! 🕊️💡
The average ROI of AI initiatives
On average, AI initiatives yield a 6% return on investment, with top performers achieving 13%. However, given the average capital costs of 10%, many AI initiatives are not profitable for the majority, as shown in a study by IBM.
Figures vary regarding the number of AI initiatives that fail to reach operational status, with estimates ranging from 50% to 80%. Thus, only 20%-50% of AI projects that do make it to operation manage to generate a mere 5% ROI on average.
Source: Generating ROI with AI (ibm.com)
This study highlights a critical insight:
While the majority of AI initiatives never launch due to various factors, the few that do were likely not the optimal AI use cases to invest in, or they failed to reach their full potential.
Such figures are certainly not for shareholder's eyes 🤓 But let’s understand these figures a bit more in detail. Understanding Capital Cost and its Implications for AI Initiatives 💸
Capital cost refers to the one-time expenses incurred when a company invests in long-term assets such as buildings, machinery, or technology. These costs are pivotal as they have a lasting impact on the financial health of a company.
A capital cost of 10% means that for every $100 spent on these assets, there is a cost of $10 associated with either the interest on borrowed money or the earnings forgone from using the company’s own funds. This 10% rate is a common threshold used across various industries to evaluate the financial viability of investments, reflecting both the burden of financing and the opportunity cost of capital allocation.
So applying this financial concept to AI initiatives as an investment just means assessing whether the money spent on these projects will generate enough returns to at least cover this 10% cost. If an AI project costs $100,000, for instance, it needs to bring in at least $10,000 just to break even on the capital cost.
This evaluation helps businesses decide if an AI project is likely to be financially worthwhile or if it could lead to a loss.
But, let's illustrate this theoretical study and figures with a hypothetical example:
A Case Study
A mid-sized manufacturing firm launched an ambitious AI initiative aimed at automating its inventory management system. The project's goal was to reduce human error, streamline operations, and ultimately save costs associated with overstocking and understocking.
Investment Details: The total capital invested in the AI project was $500,000, covering hardware, software, and personnel training costs. The project was expected to deliver at least a 10% return annually, aligning with the company’s benchmark capital cost of 10%.
Outcome: After one year of implementation, the AI system managed to improve inventory accuracy and reduce waste. However, the financial benefits calculated at the end of the year amounted to a $40,000 saving—only an 8% return on investment, falling short of the anticipated 10%.
Implications: The project’s ROI of 8% meant that it did not meet the 10% threshold required to cover the capital costs, resulting in a net financial loss.
Several factors contributed to the shortfall:
Integration Challenges: The AI system required more extensive customization than initially planned, leading to higher upfront costs.
Operational Disruptions: During the rollout phase, disruptions in warehouse operations led to temporary losses in productivity.
Adoption Barriers: Staff were slower to adopt the new system fully, limiting the expected efficiency gains within the first year.
Now the question would be for an AI Product Manager: Would you stop the initiative? AIPM Thoughts on Case Study
Well, I definitely need to find compelling arguments for the financial department, in case I choose to continue the initiative.
So let's analyze the options:
Suppose the integration challenges and adoption barriers identified can be addressed effectively. In that case, the system might have the potential to meet or even exceed the initial ROI expectations in subsequent years. It would be crucial to analyze whether these issues are inherently solvable with additional investments in training, customization, or process adjustment.
Assess the strategic importance of the AI system beyond just immediate ROI. If the system is critical for maintaining competitive advantage or is expected to provide significant operational benefits in the long run, it may be worthwhile to continue the initiative, even if it does not have immediate ROI.
Conduct a revised cost-benefit analysis incorporating the lessons learned during the first year. This analysis should include not only the financial aspects but also factors like improved accuracy, reduced waste, and potential increases in customer satisfaction.
Gather insights from the staff who use the system daily. If the staff are beginning to see the benefits as they adapt to the system, and if further training could enhance their proficiency, this could justify continuing the project.
Ultimately, the decision to continue or stop the initiative should be grounded in a thorough assessment of its potential success and how well it aligns with the company's strategic objectives.
Proceeding with the initiative might be the right call, particularly because we're just 2% shy of the benchmark. However, this fact alone may not be persuasive enough for the sponsors.
Therefore, breaking it down as I've described above could prove helpful.
💸 But here's a key takeaway
Always strive to choose AI use cases expected to yield at least a 15% ROI. The path to a successful AI initiative is unpredictable and laden with unforeseen challenges and expenses. By aiming for a minimum of a 5% buffer, you allow for a margin that can cushion unexpected costs, thus safeguarding a positive return on your investment.
Source: Failure of AI projects: understanding the critical factors - ScienceDirect
A well-defined AI Product Management function can proactively address at least four categories of challenges, anticipating and mitigating risks of failure in AI projects both before they occur and during the initiative's progression.Final Thoughts
You may already know that I am a passionate advocate for AI Product Management.
Many challenges that lead to AI initiatives falling short can indeed be addressed through the AI PM role. An AI product management team is undeniably essential for sustainable success.
However, it’s crucial to recognize that the presence of AI product managers doesn’t guarantee success for AI products, as illustrated in the example above.
Navigating the path to success with an AI product is a complex, multifaceted journey involving numerous factors:
The quality of training data,
The selection of appropriate algorithms,
Rigorous model training,
Adherence to ethical guidelines.
In addition, elements such as:
User experience,
Scalability,
Interpretability,
Security,
must not be overlooked.
Other critical aspects include regulatory compliance, feedback mechanisms, competitive positioning, and judicious resource allocation, each of which is crucial.
Indeed, also the timing of the product launch and the chosen business model have a significant impact on the final outcome.
Every one of these factors can potentially derail an AI initiative.
Quite a lengthy list, wouldn’t you agree? 😊
Takeaway for companies
Consider this: What if no one on your team assumed responsibility for overseeing the entire value chain? How likely would it be, then, that these issues are properly addressed?
Would a Data Scientist take on the responsibility for competitive positioning or a Data Engineer for the end-user experience? What about your ML Engineer's focus on ensuring the return on investment?
Yes, it's quite probable they’re not.
But even with a top-tier AI Product Team, including skilled Data Scientists, Engineers, and AI Product Managers, there's no guarantee of market success or organizational impact.
💸 But one thing you can count on: While an AI Product Manager won't promise to make you rich 💰, it guarantees you will only lose a little money! 😜Takeaway for AIPMs
While AIPMs cannot guarantee the success of an AI product or the profits it may generate, we are the ones who will alert if it’s poised to be a poor investment.
It underscores that an AIPM never has complete control, nor will we likely build unicorn products. But our role involves staying deeply connected to every facet of the AI Product Development process, grasping the technical details as necessary, and assessing the evolving needs of your customers.
Should any variables shift, we are positioned to intervene or, if necessary, clearly communicate the setbacks to our stakeholders and devise a solution - whether that means pausing the initiative to address foundational issues, adding more wo(man)power to correct the course, or just completely end that endeavor.
Isn’t it beautiful?
Isn’t it completely beyond AI?
JBK 🕊️
#AIProductManagement