AI Transformation: Is ROI the right metric to focus at this stage?

The year is 1900.?You're the CEO of the leading horse-drawn buggy manufacturer,?renowned for your craftsmanship and efficiency.?The winds of change,?however,?are blowing. The first automobiles are sputtering onto the streets,?clunky and unreliable,?but hinting at a future where horses are relegated to pastures and racetracks.?The question looms:?Will doubling down on buggy manufacturing efficiency,?perhaps by streamlining your assembly line or sourcing cheaper leather,?help you compete in this new era?

The answer (with the benefit of hindsight) is a resounding no.?The automobile isn't just a better buggy; it's a fundamentally different mode of transportation,?one that will redefine the very concept of mobility.?No amount of buggy optimization will bridge that gap.

Yet,?in the face of the AI revolution,?a similar narrative is playing out.?An increasingly large number of analysts, investors and the press are questioning the ROI of AI investments,?how to measure their success,?how to ensure they deliver tangible value.?Is this a focus on optimization,?on squeezing incremental gains from a technology that promises exponential change?

The field of finance offers a compelling lens through which to view this challenge:?the concept of "real options." Rooted in the Black-Scholes model for pricing financial options,?real options theory extends this framework to the realm of business investments.?It recognizes that strategic decisions,?particularly in innovative or volatile sectors, are rarely one-off events.?Instead,?they often unfold in stages,?with each step contingent on the outcomes of previous ones and with decision makers having the ability to react to new information. This creates a series of options - the right,?but not the obligation,?to take further action.?The value of these options lies not just in the immediate payoff,?but in the flexibility and potential they offer in the face of an uncertain future.?In essence, real options theory provides a structured way to quantify the value of "keeping your options open" in a world where the only constant is change.

Let's ground this concept in a concrete example.?Imagine you're the head of customer service at a large financial services firm.?You've been following all the news around Gen AI and have heard it has applicability in enabling AI-powered customer service agents. However, you are staring at a multi-million dollar implementation at a time when budgets are tight. Instead of viewing this as a binary "invest or don't invest" decision,?let's apply a real options lens:

1. Pilot project (the option):?Deploy the AI agent on a limited scale,?perhaps for a specific product line or customer segment.?This initial investment is the price of your option - the right to learn and gather data about the technology's performance,?customer satisfaction,?and potential ROI.
2. Gather data and learn (evaluating the option):?During the pilot,?meticulously track key metrics: customer satisfaction scores,?call deflection rates,?agent productivity,?and cost savings and its impact on returns.?While the quantitative valuation of an option is possible, it is not necessary, if we just want to use the intuition of real options as a guide. For example, the value of the option increases with increase in volatility that is when there is a lot of variability in potential outcomes and associated payoffs.? This classic McKinsey article is an excellent primer on how to use real options and a great place to start.
3. Decision Points (exercising the option):?Based on the pilot results,?you have several options:

Expand:?If the AI agent proves successful,?you can exercise your option to scale it across your customer service operations.

Pivot:?If the initial results are mixed,?you might pivot,?using the insights gained to refine the AI agent's capabilities or target it to different use cases.

Abandon:?If the pilot is a clear failure,?you can cut your losses and explore alternative solutions.?The key is that you haven't committed to a full-scale deployment until you have more information.

The challenge then lies in determining the appropriate allocation for these "growth bets." While there's no one-size-fits-all answer,?the principle of portfolio diversification offers guidance.?Just as a financial portfolio balances more stable bonds with high-risk,?high-reward stocks,?an AI investment strategy should include a mix of core AI initiatives aimed at improving existing processes and transformational AI bets that could fundamentally change your business.

The exact allocation will depend on your organization's risk appetite,?strategic objectives,?and industry dynamics. However,?it's crucial to allocate some portion of your change budget to these transformational bets, otherwise, you could be missing the forest for the trees – like the anecdotal carriage CEO we started this article with.

The real options mindset encourages you to view these transformational bets not as isolated gambles but as part of a broader portfolio.?Some bets may fail,?but the learnings from those failures can inform future successes.?And when one of those moonshots’ hits,?the payoff can be transformative,?propelling your organization to new heights in the AI-powered future.

In the beginning of this AI era,?the ability to test and learn is paramount.?Real options theory provides a framework for embracing uncertainty, making informed decisions,?and ultimately thriving in a world of constant change.