AI Business Value & Portfolio Strategy• 8 min read

Driving AI Business Value Realization: Manage Your Enterprise AI Initiatives With An Investment Portfolio Mindset

Benson Chan, Senior Partner, SOTNovember 12, 2026
Executive Takeaway

Doing AI and getting real value from it are two different things. An organization might deploy several AI solutions while struggling to demonstrate improvements in profitability, customer retention, or operating performance. Why enterprise leaders must manage AI initiatives with an investment portfolio mindset.

Enterprise AI has reached a transition point. Organizations have moved well beyond the initial curiosity and experimentation with AI. Business leaders are investing in projects and approving budgets. Technology teams are deploying applications. Business units are integrating AI into their operations. Agents are automating an increasing number of day-to-day operational activities.

The pace of AI activity is difficult to miss. The business outcomes from those activities, however, are considerably harder to see.

For many enterprises, the conversation around AI still revolves around simple activity-based metrics. How many employees are using AI? How many processes have been automated? How many pilots have reached production? How many agents are we running?

These are useful measures of progress. But they do not, by themselves, establish business value from the AI activities. They do not answer the question “What has changed since we did this?”

An organization might deploy several AI solutions, improve individual workflows and report impressive adoption figures while struggling to demonstrate a corresponding improvement in profitability, customer retention, revenue growth or operating performance.

This is a common disconnect with enterprises adopting AI. Doing AI and getting real value from it are two different things. And understanding that difference has significant implications for how enterprises plan, fund and manage their AI investments.

The problem with measuring AI "success"

Consider a company that introduces an AI assistant across its customer service operations. The assistant handles a significant proportion of routine queries, reduces the workload on support agents and shortens response times. The implementation appears successful.

But is it really?

Let’s ask the questions a CEO would ask of any investment. Has the organization reduced its cost per customer interaction? Has it improved customer satisfaction? Are support employees spending more time resolving complex issues? Has the company been able to handle higher volumes without increasing headcount? Has the investment generated enough value to cover licensing, integration, infrastructure and ongoing model consumption? And more importantly, how has these investments affected the bottom or top lines? These questions reveal whether the organization has truly made a good business investment.

A reduction in response time is an operational improvement. Its investment and the outcome depends on what the organization does with the capacity it has released. If employees remain occupied with the same workload and the business sees no improvement in service quality, throughput or operating cost, the financial return might remain limited.

The same principle applies across marketing, finance, procurement, manufacturing and other functions. Faster content production, automated reporting and improved document processing are useful capabilities. Their impact becomes clearer when they contribute to measurable improvements in business performance. The challenge for enterprise leadership is to connect these individual improvements to the impacts and outcomes the organization already cares about. Many organizations measure AI success by how well the technology works. They should measure its impact on the business and organization instead.

Business value and impact starts with solving problems

One of the most common mistakes I see in enterprise AI investments is buying the technology and then looking for a problem to solve with it. A team identifies an interesting use case, a promising platform or an automation opportunity and starts exploring its applications. The business case often develops around the capabilities of the technology rather than a clearly defined business problem.

An impact-first approach starts with the problem. The right questions establish the basis for a meaningful AI initiative. For example, a manufacturing business identifies recurring production delays caused by unplanned equipment downtime. Predictive maintenance powered by AI could help identify potential failures before they occur. But the business case should be built around the consequences of those failures: lost sales due to inability to fill orders, lost production capacity, high maintenance expenditure, and customer defections.

The technology then becomes the means of addressing a defined business problem. This approach changes how leadership evaluates proposals. An initiative with impressive technical specifications might deserve a lower priority than a relatively straightforward automation opportunity with a stronger financial case and a shorter route to implementation. AI investment decisions should be driven by the value of the business problem being addressed, rather than the novelty of the solution being proposed.

Know and manage the total cost of ownership of your AI investments

AI business cases often underestimate the full cost of ownership. The initial platform subscription or model development expenditure represents only part of the investment. Production deployments introduce recurring costs associated with model inference, API consumption, cloud infrastructure, data storage, integration, security, monitoring and ongoing maintenance. For example, AI models drift over time and must be retrained periodically to prevent false predictions, bad recommendations and compliance risks. Similarly, as user adoption grows, surging token consumption from complex prompts and longer conversations can cause operational expenses to scale faster than anticipated.

There are organizational costs too. Employees need training. Existing workflows require redesign. Data may need cleaning and restructuring. The current technology infrastructure may require modernization and upgrades. Governance arrangements must evolve as AI systems gain access to business information and operational processes.

Then there are the costs incurred from duplicating AI efforts. In large organizations, it is common for multiple departments to independently subscribe to the same AI platforms, purchasing overlapping capabilities or developing similar internal solutions. Each initiative is reasonable within its own department, but across the enterprise, the combined expenditure tells a different story. Beyond paying twice for the same work, this fragmented approach sacrifices enterprise-wide volume discounts, contractual leverage, and purchasing synergies.

This is why AI economics must be evaluated and considered at both the initiative and portfolio levels. Leadership needs visibility into the total cost of ownership, recurring expenditure, shared capabilities and dependencies across investments. It also needs to understand whether a new initiative builds on existing enterprise infrastructure or introduces another layer of complexity. A credible business case must account for the cost of getting AI into production and keeping it useful over time.

Know what business value and impact looks like before you start.

An AI solution reaching production is an important achievement. It is also the point at which the organization begins to discover whether its assumptions hold up in everyday operations. A model might perform well during testing but struggle with inconsistent production data. An automation tool might reduce processing time while introducing additional review requirements. An AI assistant might generate accurate responses but require expensive human supervision. These outcomes do not necessarily indicate a failed investment. They indicate the need to evaluate performance against the original business case.

Value realization requires a connection between technical performance and business performance. For every initiative, leadership should establish a clear set of measures before deployment. These might include:

Financial outcomes such as cost reduction, revenue contribution and improved margins.
Operational outcomes such as throughput, cycle time, error rates and resource utilisation.
Customer outcomes such as satisfaction, retention, response quality and service consistency.
Workforce outcomes such as capacity released, employee productivity and time redirected towards higher-value work.
Strategic outcomes such as faster decision-making, improved forecasting and stronger organizational capabilities.

The appropriate measures will differ across initiatives. What matters is that they have a clear relationship with the business objective.

A model’s accuracy score cannot substitute for a reduction in operating costs. Employee adoption cannot substitute for productivity improvement. A successful pilot cannot substitute for evidence of value at scale.

Manage AI as a portfolio of investments, not as a set of individual initiatives

Enterprise AI is rarely a collection of independent investments. A new customer service capability might depend on shared data infrastructure. An internal knowledge assistant might rely on the same retrieval architecture as several other applications. Multiple departments might require similar model management, security and governance capabilities.

These relationships have financial and strategic implications. When leadership evaluates initiatives individually, it risks overlooking duplication, shared dependencies and opportunities to build capabilities that benefit the wider organization.

A portfolio perspective provides a broader view. It allows decision-makers to compare initiatives based on business value, investment economics, feasibility, organizational readiness, risk and strategic relevance. It also helps determine the appropriate sequence of investments.

Some initiatives offer immediate operational improvements. Others require foundational capabilities that make future deployments easier, safer or more economical. Certain investments might be worth pursuing for their strategic potential, even when their financial returns take longer to emerge.

The objective is to allocate capital and organizational resources across the portfolio in a way that balances near-term returns with longer-term capability building.

This requires ongoing review. Business priorities change. Technology costs evolve. Initial assumptions prove inaccurate. Some initiatives deliver stronger results than expected, while others struggle to move beyond limited adoption. A portfolio that remains unchanged despite new evidence is unlikely to remain aligned with business priorities.

Value realization requires ownership and accountability

There is another dimension that often receives insufficient attention: who owns the value after an AI solution goes live? Technology teams might own the platform and its technical performance. Business functions might own the processes being transformed. Finance might be responsible for validating financial benefits. Risk and governance teams might oversee compliance and controls.

Without clear ownership, the organization risks creating a situation where everyone contributes to deployment but nobody is accountable for the business outcome.

Value realization requires defined responsibilities.

The business function should own the outcome. Technology teams should ensure the solution performs as intended. Finance should establish credible measurement methods. Leadership should review whether the investment continues to justify its cost and resources.

This accountability must continue beyond implementation. A solution that initially delivers strong results might lose value as business conditions change, model costs increase or employees develop workarounds. Regular reviews help identify these changes and determine whether the organization should improve, expand, redesign or discontinue the investment.

AI governance, therefore, has a direct relationship with business value. It helps ensure that investments remain aligned with business priorities, operate within acceptable boundaries and deliver outcomes that leadership can evaluate.

Sustained enterprise AI success requires an investment discipline

The next stage of enterprise AI will bring greater technical capability, more autonomous systems and an expanding range of potential applications. The number of opportunities available to organizations will continue to grow. The harder task will be deciding which opportunities deserve attention, funding and organizational capacity.

Enterprises will need to become more deliberate about the relationship between AI strategy, investment economics, execution and measurable outcomes. They will need to understand what they already own, what they are building, what each initiative depends on and where the next investment is likely to create the greatest value.

This calls for a more connected approach to AI investment management.

At Strategy of Things (SoT), this is the problem we believe enterprise leadership needs to address. Through SoT AI Portfolio Compass™, we are building an Executive AI Operating System that brings together AI initiatives, investment economics, business value, readiness, dependencies and governance within a structured decision-making environment.

The aim is to give leadership a clearer basis for evaluating opportunities, prioritising resources, understanding trade-offs and deciding what to fund, scale, defer or stop. Ultimately, AI business value is established through the decisions surrounding the technology: the problems an organization chooses to solve, the investments it prioritises, the capabilities it builds and the outcomes it holds itself accountable for delivering.

The measure of enterprise AI progress will not be how much AI an organization deploys. It will be how much meaningful business value it creates from the investments it makes.

Discuss Your AI Portfolio with Our Executive CAIO Practice

Align AI investments, prioritize high-impact use cases, and mitigate operational risk.

Book Discovery Session