Site icon Evangelos Simoudis

The AI GPS: A Dashboard for the Enterprise CFO

Evangelos Simoudis
Sample CFO AI dashboard

 Many corporations are starting their AI journey in the wrong way. The result is what I previously called The AI Pilot Purgatory. In the process, they make two mistakes. First, they lack an AI strategy. Second, they don’t engage the right members of their senior management teams in their AI-related decisions. As they face skyrocketing AI charges, CFOs find themselves in the middle of a tug-of-war among the corporation’s various organizations. In this post, I describe a four-graph dashboard that acts as the corporation’s AI GPS. The CFO owns the dashboard. The executive team uses the dashboard to monitor the corporate AI strategy’s performance, and the CEO uses it to communicate the effectiveness of the corporate AI efforts. 

Moving without GPS

For the past two years, legacy corporations have reacted to the AI Stakeholder Squeeze (the simultaneous AI-attributed pressure from boards, investors, and employees) by launching Centers of Excellence and funding random AI pilots. Recent research by PwC found that most companies are seeing no financial return at all from their AI efforts.

Fifty-six percent of the 4,454 CEOs surveyed report neither higher revenues nor lower costs from AI over the past twelve months, and only one in eight report both. With an estimated $1.4 trillion in AI capital expenditures projected across hyperscalers by 2028, institutional investors are questioning AI’s Return On Invested Capital (ROIC). Economists have recently pointed out that much of the current economic activity is driven by a speculative infrastructure build-out, while actual firm-level productivity gains remain scarce. As this massive infrastructure bill comes due without measurable enterprise returns, the AI Stakeholder Squeeze is about to become exponentially worse.

PwC’s reading of the data points in the same direction. Isolated, tactical projects do not produce measurable value, and returns come only from deployment at enterprise scale that is consistent with business strategy. This is a restatement of Pilot Purgatory from the auditor’s side of the table.

As it attempts to lessen the Squeeze, the corporation needs an AI strategy and a GPS to help it stay on the right path while executing the strategy. Forging an AI strategy and being responsible for its execution requires bringing together the corporation’s CFO, CIO/CTO, head of HR, Business Unit Leaders, and Chief Legal Officer. The strategy must be created by these corporate executives and not a consultant. (A consultant may facilitate the activity). But once that strategy is established, the team needs a “GPS” to monitor its execution, make certain it is moving in the right direction, and adjust it if necessary. Corporations must avoid reactions such as those observed recently with tokenmaxxing and tokenminimizing.

When a corporation first applies AI to a process, it inevitably hits a Dip. During the Dip, costs spike, existing workflows break, and employees’ anxiety increases. The executives must understand and explain the Dip. Otherwise, their AI strategy would fall apart before it is implemented. For this reason, the team that owns the strategy must have access to the right data connecting AI-related costs with outcomes and derive the right insights from it. We have determined that the four-graph dashboard presented next provides the foundation for creating these insights. 

The CFO owns this dashboard. In this way, we eliminate the “shadow IT” and “shadow HR” budgets that traditionally plague the transformation efforts of legacy corporations.

The Four-Graph Dashboard

The dashboard relies on four interlocking graphs, each owned by a specific executive but monitored collectively by the cross-functional team that owns the strategy and its execution. Three of them are sensors and one is a roll-up. Adoption is the leading indicator: it moves first and predicts what the other curves will do. Productivity is coincident: it reports what is happening now on the floor. Investment is the control variable the executive team can act on directly. The Portfolio J-Curve lags all three because it aggregates them. A GPS is useful only because it describes the road ahead rather than the road already traveled, and the same is true here. A team watching only the Portfolio curve is navigating by rear-view mirror.

  1. The Adoption S-Curve. Technology does not produce ROI; changed human behavior does. To achieve true ROI, the corporation must create an internal network effect. This graph tracks the rate at which the retrained workforce uses the reimagined AI-centric workflows. It organizes these employees from early adopters to laggards. In the process, it shows which employees require additional training, which will not be able to make the transition to the new process, and where the process needs to be adjusted. The adoption accelerates as culture shifts, and eventually plateaus at full adoption of the AI-centric process is achieved. The curve signals to the HR organization when it needs to intervene for the corporate transformation to continue. 
  2. The Transformation Investment Inverted J-Curve. This graph tracks the fully burdened, upfront cost of the enterprise’s AI transformation. In addition to the corporation’s AI-related technology spend, e.g., cloud infrastructure and frontier model API calls, it captures the operational costs of redesigning legacy business processes and HR’s budget for workforce upskilling. If a corporation only tracks its IT bills, it obscures the true depth of the financial dip. Without such comprehensive telemetry, the CFO cannot accurately evaluate the unit economics associated with the AI-driven transformation. This curve is inverted because the intensive, front-loaded investments spike during the initial rollout, but should drop significantly over time as new processes normalize and infrastructure scales. 
  3. The Productivity J-Curve. This graph measures operational friction and throughput by AI project. It tracks performance metrics like error and rework rates, straight-through processing percentages, and the fully burdened cost per successful outcome. The Investment curve reports transformation spend in aggregate. The Productivity curve divides the operating portion of that spend by successful outcomes to expose unit economics. The executive team should confirm the allocation rules once and then stop relitigating them. During the initial period of using the AI-centric process, throughput drops and unit costs rise. Employees, starting with the early adopters, learn the process, verify and correct problematic outputs, and manage broken hand-offs. The Dip may last longer and be deeper in a legacy corporation than in a digital native. If the work is done properly, then the Dip is not the result of AI’s failure, but the result of the new processes replacing ossified ones. 
  4. The Portfolio J-Curve. The projects established by the corporate AI strategy to achieve the corporation’s goal, or goals, must be managed as a portfolio. This graph aggregates the data from the other three curves into a single view. It subtracts the massive upfront costs (the Inverted J-Curve) and accounts for the initial operational drag (the Productivity and Adoption curves) to map the enterprise’s timeline for achieving the established goals. By charting the net financial return, this curve justifies the macro-level capital expenditure to the board in a predictable timeline. 

One reading convention prevents most misinterpretation at a glance. On three of the four curves, up is good. On the Transformation Investment Inverted J-Curve, up is the cost of the transformation and down is progress. Ownership follows the same logic as the strategy team itself. The CIO/CTO owns the Investment curve, the Head of HR owns the Adoption curve, the relevant Business Unit Leader owns the Productivity curve for each project, and the CFO owns the Portfolio curve along with the dashboard as a whole. Ownership here means budget authority, not reporting responsibility. That distinction is what eliminates the shadow IT and shadow HR budgets, because an executive who must fund a line item can no longer route around it.

Operationalizing the Dashboard

Creating the dashboard is half the battle. The executive team that owns the strategy must derive the insights that give it confidence the AI projects are proceeding on plan, provide it with the early warning to intervene when they stumble, and enable the CEO to remain deeply involved with the AI effort, while keeping the board and external stakeholders properly apprised.

Setting the baseline

Before the first project starts, the executive team must commit to an expected depth and duration for the Dip on each initiative, and record it. This is the single most consequential act of the strategy process, and the one most often skipped. Without a baseline, every subsequent month of poor numbers is a matter of opinion. The CFO argues the project is failing, the BU Leader argues it needs more time, and the argument is settled by whoever has more political capital rather than by evidence.

The baseline need not be precise to be useful. A trough depth expressed as a percentage decline in throughput, a duration expressed in months, and a stated recovery threshold are enough. Legacy corporations should expect deeper and longer Dips than digital natives, and should set their baselines accordingly rather than importing benchmarks from companies that never had to dismantle a thirty-year-old process. What matters is that the number exists in advance and that overrun against it becomes a tripwire rather than a debate.

The internal cadence

Quarterly governance is too slow to manage AI transformation. The executive team must operate on a dual-track cadence to manage velocity without causing executive burnout:

Dip or death spiral

The size of this problem is measurable. In PwC’s survey, twenty-two percent of CEOs report that AI has increased their costs, and roughly thirteen percent of all respondents report higher costs with no revenue gain whatsoever. Every one of those companies is somewhere on a curve. Some are in Month 6 of a healthy Dip on a process that will produce a step change in throughput. Others are funding an expensive failure. Nothing in the survey distinguishes them, and in most cases nothing in their own reporting does either. The dashboard closes this gap. The executive team states that the Dip is a symptom of transformation rather than of failure.  

The CEO’s narrative engine

The CEO uses the dashboard to remain deeply involved in the AI strategy and to speak about it with precision to three audiences. The dashboard converts surprises into variance reports. A CEO who has publicly committed to the expected depth and duration of the Dip reports a difficult month as being on plan. A CEO who has committed to nothing reports the same month as a setback. The credibility is not identical.

Conclusion

The AI Stakeholder Squeeze is not a temporary phase. The announced AI investments and the budgetary tug-of-war unfolding inside corporations will only increase the forces at play.

The consequence of operating without instrumentation is specific and worth stating plainly. A healthy Dip and a failing project look identical from the outside: costs up, throughput down, employees unhappy. A corporation that cannot tell them apart will make systematic errors. It will kill its most ambitious projects. The projects that touch the most ossified processes and therefore dip the deepest. Conversely, it will protect its shallowest ones, because those look better on a quarterly review. Two years of this produces a portfolio of thin automations, a large bill, and a board that has concluded AI does not work here.

The dashboard is uncomfortable by design. It makes the depth of the trough visible and attributable. A CFO who charts that trough owns it in a way that no one owned it. That discomfort is not a flaw in the instrument. It is the price of being able to defend the strategy with evidence rather than conviction. It is considerably cheaper than the alternative.

Corporations with an AI strategy need a GPS to monitor it, derive insights from it, adapt it when the terrain changes, and communicate its results to internal and external constituencies. The four-graph dashboard provides these capabilities.

Exit mobile version