7 Factors That Drive Returns on AI Investments - And Why Your CFO Should Be in the Room

This sounds obvious, but the survey revealed a telling disconnect. About 14% of respondents said they're getting a great deal of value from AI but only slight ROI. Another 9% reported moderate value but substantial ROI. Those two groups are clearly measuring different things.

The researchers put it simply: value should be understood as "AI is doing what we want it to do." Some organizations pursue short-term financial returns. Others are playing a longer game, building a technology foundation that positions them for future transformation. Both approaches can work, but you have to pick one and be explicit about it.

For a Houston-area business, the practical question is: what does "working" look like for your AI spend? If you're a CPA firm using AI to speed up document review, the metric might be hours saved per engagement. If you're a construction company using AI for project estimation, the metric might be bid accuracy or win rate. The worst position to be in? Spending money on AI tools without a clear answer to "what are we trying to accomplish here?"

Most organizations default to using AI for internal process improvements - automating repetitive tasks, speeding up data entry, generating internal reports. That's where most businesses start and it makes sense. But the survey found that high-value organizations are also focused on embedding AI into their customer-facing products and services.

The study highlighted executives at companies like Schneider Electric, where internal AI delivers more immediate financial returns while customer-facing AI represents a longer-term play focused on capturing market share. Each requires different approaches to measuring success and different timelines for seeing results.

There's also a defensive angle that doesn't get enough attention. One medical device executive made the point that AI capabilities in products protect market share - customers expect it, and companies that don't offer it will lose business to competitors who do. That's not a revenue gain. It's a revenue preservation strategy. And it's just as valid a reason to invest.

For SMBs in Houston and Katy, think about this in two tracks. Track one is the obvious one: where can AI make your internal operations faster or cheaper? Track two requires more imagination: can AI improve the service you deliver to clients? A law firm using AI to deliver faster research to clients, or a wealth management firm using AI-powered analytics in client presentations - those aren't just efficiency plays. They're competitive differentiation.

This finding should make every business owner pause and reconsider their assumptions. When asked which type of AI produces the most value, 50% of survey respondents pointed to analytical AI - things like dynamic pricing, customer targeting, and predictive analytics. Rule-based AI came in a close second at 40%, covering applications like insurance underwriting, anti-money-laundering systems, and robotic process automation. Generative AI? Just 9%. Agentic AI came in at 2%.

Generative AI dominates every news cycle and conference stage. It's the category most people think of when they hear "AI." But for businesses focused on actual returns, the older, less glamorous forms of AI are pulling significantly more weight.

That said, the study did find that companies adopting agentic AI - the newest category, where AI systems can take independent actions - are 22% more likely to report achieving a great deal of overall AI value. Agentic adoption appears to be an indicator of organizational maturity, even though the technology itself isn't yet the primary value driver.

Companies that extract real value from AI tend to have a structured process for moving initiatives from idea to production to measured outcomes. The survey found multiple approaches that work: one financial services company uses a custom "AI playbook" that guides business lines from use-case exploration through responsible deployment. An electrical utility applies a stage-gate method borrowed from R&D. The most common approach among the executives interviewed was a digital product orientation - managing each AI initiative as a product with defined stakeholders, timelines, and accountability.

One executive credited the product orientation as the single most important factor in achieving value, because it creates a structure for proposing expected benefits, reviewing results over time, adjusting the approach, and holding stakeholders accountable for what gets delivered.

The framework has to include a data readiness component too. The survey found that 55% of respondents cited unready data as an obstacle to getting value from AI. That's consistent with other research we've seen. You can have the best framework in the world, but if your data is scattered across disconnected systems with no quality controls, AI can't perform.

This is maybe the most striking finding in the entire study. Most organizations assign AI value accountability to their chief data/analytics/AI officer (38%) or individual functional executives (35%). Only 2% assign it to the CFO. But when the CFO is responsible for achieving AI value, 76% of organizations reported getting a great deal of value from their investment. Compare that to 53% under CIOs or CTOs, and just 32% under functional line executives.

Why the massive gap? Finance leaders bring measurement rigor, organizational credibility, and the authority to hold other departments accountable for results. They're trained to validate claims, question projections, and demand evidence. When the person asking "did this AI project actually deliver?" is the same person who controls the budget, the answers tend to get a lot more honest.

The study highlighted DBS Bank in Singapore, where the finance organization partners with technology executives to certify AI value. Each business unit's CFO validates their respective results, which then roll up to an aggregate company-wide number that gets published in the bank's annual report.

For a small business, this doesn't mean hiring a CFO just for AI. It means whoever manages your books - your controller, your outsourced accounting firm, your bookkeeper - should be part of the conversation when you're evaluating AI results. Is the tool actually saving the hours we projected? Did the automation reduce the cost we expected? If nobody with financial discipline is asking those questions, the answers tend to be vague and self-serving.

The training gap runs in two directions, and both matter. According to the survey, 58% of organizations have not trained their employees on AI productivity tools and workflows. Separately, 29% acknowledged that their leadership team lacks sufficient understanding to drive AI value creation. Organizations that invest in both employee skills and leadership fluency see a 23-percentage-point advantage in value realization compared to those that don't.

Here's the detail that changes the narrative: only 13% cited workforce resistance as a barrier to AI value. Employees aren't pushing back. They're waiting. Waiting for clear direction from leadership, waiting for training that makes the tools useful in their specific roles, waiting for someone to remove the other barriers like missing frameworks and messy data.

The researchers developed a maturity model based on three components: putting AI into actual production (not just running pilots), assessing the value of those production use cases before and after implementation, and aggregating and reporting that value across the organization. Based on survey responses, they identified six stages with dramatically different outcomes.

The stages and the percentage of organizations in each that reported getting a "great deal" of AI value tell a clear story:

• Stage 0 - Unmeasured Pilots (3% of organizations): Running AI experiments with no outcome measurement. Only 4% in this stage report high value.
• Stage 1 - Production Without Assessment (11%): AI is in production but nobody's evaluating the business impact. Just moving past pilots jumps the number to 18% reporting high value.
• Stage 2 - Pre-Implementation Assessment (17%): Building ROI projections and business cases before deployment, but not validating outcomes afterward. Only a modest improvement to 20%.
• Stage 3 - Post-Implementation Assessment (30%): Measuring individual use cases after deployment. This is the first major inflection point - 44% report high value, more than double Stage 2. But companies seem to get stuck here, spending a median of six years at this stage.
• Stage 4 - Aggregated Annual Assessment (21%): Rolling up AI value across the portfolio annually and making results available internally. Another significant jump to 58% reporting high value.
• Stage 5 - Formal Reporting (16%): Reporting AI value to boards, investors, or public markets. This is the second and largest inflection point - 85% report high value.

Two things jump out. First, the biggest single improvement comes from measuring AI outcomes after deployment - not before. Business cases and projections are fine, but the real value signal comes from asking "what actually happened?" Second, organizations that formally aggregate and report AI value see dramatically better results, likely because the discipline of reporting forces accountability and honest assessment.

For a 50-person business in Sugar Land or Cypress, you're not going to publish AI results in an annual report. But you can absolutely track what your AI tools deliver, review those numbers quarterly with your leadership team, and compare total AI spend against measurable outcomes. That's Stages 3-4, and the data shows a massive difference in results.

CinchOps is a managed IT services provider based in Katy, Texas, serving small and mid-sized businesses across the Houston metro area. CinchOps specializes in cybersecurity, network security, managed IT support, VoIP, and SD-WAN for businesses with 10-200 employees.

This survey confirms what we've seen across 30 years of technology deployments: the tool is never the hard part. Getting it to work inside a real business - with messy data, untrained staff, and unclear goals - is where most investments succeed or fail. Here's where we help Houston SMBs with their AI strategy:

• Data infrastructure readiness: We audit your data environment, identify gaps in quality and connectivity, and build the foundation that AI tools need to perform. 55% of organizations in this survey said unready data is their biggest AI obstacle - our managed IT team fixes that problem before it costs you money
• Cybersecurity for AI implementations: AI tools create new data flows and new attack surfaces. We ensure your AI adoption doesn't become a security liability
• Business process automation: We help design and implement the structured frameworks this survey identifies as a key value driver - connecting AI tools to actual workflows with defined outcomes
• Managed IT support for AI environments: As you scale AI across your business, we keep the underlying infrastructure stable, secure, and performing at the level these tools demand
• CTO/CIO advisory services: We provide the strategic technology guidance that helps you define value targets, select the right AI tools, and measure results with discipline

The path to AI value is a management challenge, not a technology challenge. We've been helping Houston businesses work through exactly that kind of challenge for decades. If you're spending money on AI and not sure what you're getting back, that's a conversation worth having.