Sanjay Dhawan, CEO at SymphonyAI. 30 years leading global tech companies. Engineer by training, operator by practice.
AI is scaling rapidly across enterprises, but the way its impact is measured has not kept pace. Companies are investing heavily, yet many still struggle to answer a straightforward question: What did we actually get from it?
We see the same challenge internally. At my company, we spend a hefty amount per month on AI coding tools to improve how we build and deliver software. The expectation is not incremental productivity gains, but measurable impact. Are we delivering faster? Are we reducing cost? Are we enabling work that was not previously possible? Those are the questions that matter, and they are often the hardest to answer with precision.
As spending continues to grow, expectations are shifting from potential to proof. Gartner projects global AI spending will surpass $2.5 trillion, and at that scale, the gap between investment and measurable value becomes harder to ignore. That gap is exposing a fundamental disconnect between how AI is priced and how it creates value.
The Gap Between Spend And Value
Most AI pricing models still reflect how software was sold in the SaaS era, not how AI delivers impact. Subscription models, typically priced per user, measure access. Consumption-based models measure activity. Neither, on their own, measures outcome.
AI creates value only when it is embedded into real workflows and produces measurable improvements in how those workflows perform. Without that connection, usage can scale without delivering meaningful business impact. To bridge this gap, organizations must shift their focus from activity to impact, moving the conversation from how much AI is used to how it improves business performance.
Defining Success Before Deployment
One of the most common reasons AI initiatives struggle to demonstrate value is that organizations begin implementation before defining what success looks like. Many teams fall back on tracking adoption metrics like the number of users or prompts, or total token consumption. Those can help measure engagement, but they rarely explain whether the business is actually performing better.
A more effective approach is to isolate a small number of operational metrics before deployment begins. In financial services, that may mean reducing investigation cycle times. In manufacturing, it may mean improving throughput or yield. In retail, it could involve increasing forecast accuracy or reducing waste. The specific metrics vary by industry, but the principle is universal: Define the target business outcome first, then measure the AI strictly against that metric.
Where Organizations Often Get Stuck
Defining outcomes up front is easier than attributing them. Most deployments influence multiple teams, systems and decision points across the organization. When performance improves, it can be difficult to determine exactly which changes drove the result.
Another common mistake is relying on objectives that are too broad to measure effectively. Goals like “improve productivity” or “increase efficiency” sound reasonable, but they often lack a clear baseline. Achieving strong results means treating measurement as an ongoing discipline rather than a final milestone, continuously evaluating progress against preestablished benchmarks.
Accountability Matters
As AI becomes more deeply embedded in business operations, expectations are changing accordingly. Enterprises are no longer evaluating AI solely on technical capability or adoption. Increasingly, they want evidence that it is improving operational performance.
That shift will require a different level of accountability from both vendors and internal teams. Organizations must establish clear measurement frameworks, while technology providers must be prepared to align around shared definitions of success. The market is moving toward a standard where value must be defined, delivered and measured with greater precision.
From Experimentation To Results
The organizations that I believe will create the most value in the AI era will be the ones that establish clear definitions of success, measure outcomes consistently and hold themselves accountable for results. Over time, that level of accountability will become the expectation, not the exception. As AI spending continues to grow, the conversation will increasingly shift from what AI can do to what it actually delivers.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
