How to Balance GenAI Costs with Business Value

GenAI is everywhere. It’s in the headlines, it’s in boardroom conversations, and it’s on every technology roadmap. But amidst all the hype and hustle around what GenAI promises, here’s an interesting reveal… many organizations end up experimenting with GenAI. They often fail to realize business value or figure out true costs in achieving GenAI use cases.

You can’t afford to adopt GenAI blindly. Just because the technology is powerful doesn’t mean that it’ll automatically drive in profits. And that’s where most GenAI initiatives fail. Calculating both value and cost upfront should always be your first step. You need a systematic approach to evaluate which GenAI use cases actually move the needle for your business and which ones are just expensive experiments.

How about a proven framework that addresses how to assess business value, calculate true costs, and prioritize use cases that deliver measurable returns?

In this guide, you’ll learn about the three categories that major GenAI use cases can be clubbed into, understand the real cost components most enterprises overlook, and get a step-by-step framework for prioritizing your AI initiatives. By the end, you’ll know exactly which GenAI projects deserve your budget, and which ones belong on the back burner.

Is Improving Individual Productivity the Only Focus of GenAI Initiatives?

Here’s the first thought that comes to mind when we talk about GenAI… it’s just about improving the productivity of employees. Write emails faster, generate code quicker, and create presentations in minutes instead of hours. That’s thinking small.

Earlier adopters across industries and business processes are reporting a range of business improvements that vary significantly by use case, job type, and skill level of the worker. Yes, the value is in individual productivity gains. But it’s also about how GenAI can transform entire business processes, create new revenue streams, and even reshape entire industries.

This is where Gartner® framework comes in handy. Instead of viewing all GenAI use cases at once, you can strategically categorize them based on their business impact and transformation potential on a holistic level.

Broad Classification of Enterprise GenAI Use Cases

As per Gartner, “the primary focus of enterprise GenAI initiatives is on individual productivity, but with notable focus on other areas”. Here’s our understanding of the “other areas”: Defend, Extend, Upend.

1. Defend

These use cases augment employees with general productivity tools. They are relatively easy to buy (not necessarily low cost) and to deploy at scale, and result in minimum competitive advantage. They will be the baseline standard and therefore have a defensive competitive impact. Without headcount reduction, you can measure the benefits of these use cases in terms of how employees use their newfound time, or “return on employee.” Based on responses to the 2024 Gartner AI Survey, 50% of enterprise GenAI initiatives are focused on the “defend” use cases.

2. Extend

These use cases extend an existing process with GenAI capabilities for potential differentiation in existing markets. Examples include redesigning customer service functions to work with conversational GenAI interfaces or AI agents, or automating key business processes. Because you can tie the result to specific revenue and cost improvements, you can measure these use cases through direct return on investment. Based on responses to the 2024 Gartner AI Survey, 30% of enterprise GenAI initiatives are focused on “extend, operations/process-level use cases.”

3. Upend

These use cases aim to transform an industry. They are characterized by the creation of new markets, game-changing products, revenue streams, business models and core processes. Examples include life-changing drug discoveries and/or enterprise value proposition redesign. The success measures are more strategic (e.g., size of new market created or amount of revenue generated from new products), and the results are realized longer term. Based on responses to the 2024 Gartner AI Survey, 20% of enterprise GenAI initiatives are focused on the “upend” use cases.

Each category requires different investment approaches, risk tolerance levels, and success metrics. Now, let’s go through some market trends first. And then show you how Gartner segments common GenAI use cases based on Defend, Extend, and Upend.

Considering AI Market Trends and Categorizing GenAI Use Cases

Understanding the latest market trends and what AI leaders really need is crucial here before jumping onto anything else. As per Gartner:

GenAI vendors are updating business models and refining deployment approaches to reduce costs and improve accuracy. This opens more cost-effective “build” options for enterprises with in-house AI expertise to boost ROI. Gartner predicts that by 2028, 30% of GenAI pilots that move forward into large-scale production will be built versus bought (i.e., deployed using packaged applications) to lower cost and increase control.

At the same time, new AI agents and low-code/no-code platforms offer easier “buy” options that require fewer skills but come with higher licensing costs. AI leaders must weigh the trade-offs between “build” and “buy” across cost, risk, security, control, and skills.

Also, there’s an interesting pattern that Gartner notes. Many organizations pilot with “buy” options for speed but later shift to “build” models to scale more cost-effectively and retain control over strategic capabilities.

What these findings indicate is that without a clear framework it’s difficult to assess which use cases are worth scaling and which ones should be quietly shelved.

To begin with, here’s how Gartner categorizes common GenAI use cases under Defend, Extend and Upend. As Gartner says, “It focuses on eight of the most common GenAI use cases that Gartner clients have invested in”.

Building on this, let’s walk you through a framework developed by Gartner that makes things easier, especially for AI leaders.

The Framework to Calculate GenAI Cost and Business Value

Let’s talk about a framework proposed by Gartner to evaluate GenAI costs and benefits. Comparing the price of different AI tools is not all. It’s more about understanding the total cost of ownership and the full spectrum of business value. Here’s our take on Gartner framework to calculate GenAI costs and value:

1. Define Your AI Ambition and Risk Appetite

The journey to realizing GenAI’s value starts with clarifying your organization’s AI ambition. Decide whether you aim to Defend, Extend, or Upend your competitive position. This ambition shapes strategy, priorities, and investment decisions.

I. Focus on New Currencies of GenAI

GenAI requires a greater acceptance of indirect or delayed returns rather than immediate ROI. Each ambition delivers its own “currency of value”:

• Return on Employee for Defend
• Return on Investment for Extend
• Return on the Future for Upend

Much of GenAI’s benefit shows up in leading indicators such as productivity, faster cycles, customer experience, quality, and workforce upskilling. These eventually reduce costs, grow revenue, or lower risk, but not usually through quick headcount or cost cuts.

You must be prepared to harvest financial benefits over time by doing more with fewer resources, using less external support, retaining customers, or reinvesting saved time into sales.

However, many CFOs expect hard-dollar results within a year, often pushing funding toward tactical projects instead of strategic, long-term initiatives. Gartner 2024 AI Mandates survey shows this tension clearly. They say that Less than 30% of respondents report that their CEOs have exceptionally high praise for the returns on their AI technology investments, exceeding expectations., while high-maturity organizations typically evaluate the success of AI initiatives through multiple metrics, including financial and business analysis, as well as technical metrics.

II. Think Like a Venture Capitalist

For transformative GenAI initiatives, leaders should adopt a VC-style approach, focusing on market-level and competitive impact, and not just short-term ROI. Metrics such as the size of new markets created or percentage of revenue from AI products are more suitable here. In some cases, entirely new metrics may be required, just as cloud adoption gave rise to annual recurring revenue (ARR) as a measure of success.

Ambitious AI goals demand greater risk tolerance and investment decisions that prioritize long-term strategic outcomes.

Here’s what experts at Gartner say:

II. Invest on People and Process Changes

Technology alone cannot deliver value. People and process changes are just as critical. Success requires early alignment with HR, finance, compliance, and strategy teams to:

• Redesign and transform work
• Manage change
• Strategically use time savings from productivity improvements
• Minimize risk from negative impacts of AI

Gartner research shows organizations that are successful at realizing expected value from GenAI spend as much or more on business change and transformation as on technology for work, process, and behavioral change management and user enablement. Scaling also depends on repeatable AI engineering practices and defining new competencies, such as prompt engineering or validation roles. Balancing human strengths with AI capabilities is key here.

III. Consider Model Costs to Be Uncertain and a Barrier to ROI

While model costs are dropping rapidly, they remain unpredictable. Token pricing, GPU rentals, and open-source models (LLMs) offer more affordable options, but hidden costs like data readiness, integration, governance, and ongoing training still challenge ROI.

Experiments may be cheap, but scaling often exposes overlooked costs in infrastructure, security, and workforce transformation. Investments in AI-ready data are critical for reducing token consumption and improving accuracy. Continuous monitoring of costs, pricing models, and market innovations is essential to sustaining value as deployments grow.

2. Defend and Strengthen Your Baseline First

Most early GenAI use cases fall into Defend and Extend categories, with Defend focusing on productivity tools that become baseline standards. These are easy to buy and deploy, but they deliver limited competitive advantage. Their value is measured as “return on employee”.

I. Measure Benefits in Terms of “Return on Employee”

Defend use cases like coding assistants or business productivity tools are often measured by time savings. But translating time savings into financial benefits is challenging. Most enterprises don’t see direct cost reductions, instead channeling gains into outcomes that influence future value such as quality, customer experience, and employee retention.

II. Manage Productivity Leaks

Not all saved time translates into business value. Productivity leaks (when time is lost to distractions, bottlenecks, or low-value work) can undermine the impact. Hence, you must actively manage how employees reinvest freed-up time in higher-value activities.

III. Link Productivity Tools to Revenue Outcomes

The financial impact of general productivity tools is hard to quantify, similar to word processors or spreadsheets. As a result, abandonment rates can be high when benefits aren’t tied to specific processes or measurable outcomes. Value increases when tools support revenue-linked goals like faster loan underwriting, improved recruiting, or customs automation.

The true return depends on factors like data readiness, change management, training, user adoption, and integration with core processes. Note that costs also extend beyond licenses to include data management, security, governance, and training.

3. Extend Existing Processes to Differentiate, Scale, and Deliver ROI

Extend use cases are where you apply GenAI to improve your existing processes. Unlike Defend use cases, the ROI here is much more measurable. Instead of just boosting productivity, you’re using enterprise data in smarter ways that give you a real competitive edge. Consider these for example:

• Using GenAI-assisted customer support to cut resolution times and reduce agent turnover
• Generating personalized sales content to lift conversion rates, increase deal sizes, and grow revenue per rep
• Accelerating document search and summarization so employees spend less time hunting for answers and more time on high-value work

When it comes to deploying Extend use cases, you’ve got options:

• Buy add-ons to platforms like Salesforce or SAP. This is cheaper for pilots but expensive to scale because of license and usage-based costs.
• Build custom applications on APIs using your enterprise data. It gives you more control, but you’ll need technical expertise.
• Fine-tune pretrained models with your own data, especially open-source models hosted in your cloud or on-premises. This helps you optimize both accuracy and cost.

Extend use cases generally require massive upfront investments when compared with Defend use cases. You should leverage rapid prototyping with APIs, optimize for specific hardware, and balance trade-offs between cost, latency, and quality.

If you manage costs well and tie these use cases directly to business processes, Extend initiatives can deliver strong financial returns. The key is aligning them with outcomes you can measure clearly.

4. Upend the Industry with Transformative GenAI Use Cases

Upend use cases are the bold moves… the ones that aim to transform industries with new products, services, and business models. They’re expensive, risky, and long-term, but they’re also where a unique competitive advantage is born. For example:

• Insurers fine-tuning models with policy data to reinvent underwriting
• Financial services firms training models on proprietary data to unlock new revenue streams
• Companies like Intuit embedding AI across products to redefine customer experience

Many organizations are testing domain-specific models, which often outperform big general-purpose ones at lower cost. But be careful. Open-source options can bring hidden costs, like the need for more skilled resources, weaker security, and future technical debt.

Here, you need to evaluate investments with strategic metrics, not just short-term ROI. Look at market share impact, revenue from AI-driven products, or the size of new markets you could create. These initiatives demand patience and higher risk tolerance.

Recommendations to Balance GenAI Costs and Value

To really capture the true value of your GenAI use cases, you need a structured, end-to-end approach for managing your entire GenAI portfolio. That means systematically ranking, testing, measuring, and scaling use cases, while building the governance, skills, and processes required for long-term success. A report by Gartner lists four “end-to-end recommendations” to calculate GenAI costs and business value, and here’s our take on those recommendations:

1. Rank, Prioritize, and Pilot Strategically

Start by collecting use cases from across your organization. Explore employee ideas. Parallelly, collaborate with senior stakeholders and strategy teams. Make sure they are in sync with your AI ambitions. Then rank and prioritize on the basis of potential impact, urgency, cost, and risk.

Build a tiger team of executives, strategy leaders, and business/tech experts to evaluate new GenAI-based products, services, and business models that could transform your industry. Communicate with your board and leadership about the need for higher risk tolerance and new investment criteria that look beyond short-term productivity gains.

Launch low-cost pilots and aim for high-value opportunities. Simulate best- and worst-case scenarios for costs and benefits to account for uncertainty. As Gartner says:

2. Precisely Calculate Costs and Value

For starters, don’t expect GenAI savings to show up automatically. Build processes to measure both financial and nonfinancial success metrics for every initiative. These should include strategic benefits that influence financial outcomes, like improved customer experience or reduced churn.

As you roll out pilots and scale, refine your tracking by:

• Measuring how employees use freed-up time and converting that into financial impact
• Embedding GenAI cost/benefit tracking into your financial operations and cost management processes
• Running A/B tests to validate GenAI’s effect on outcomes
• Using automated productivity metrics inside existing enterprise apps and development platforms
• Running scenario analysis to test the limits of licensing, tokens, and inference costs as usage grows

Also, factor in total ownership costs: model management, risk/security systems, data repositories (like vector databases), knowledge architecture, labeling, and training. Keep in mind that GenAI still depends on high-quality data. You’ll also need new governance and security capabilities, which should be tested during beta releases and phased rollouts.

And don’t forget to keep an eye on the market. Vendor pricing models change quickly, so regularly reassess whether your current deployment approach is still the most cost-effective one.

3. Measure Only What Matters

Tracking outcomes shouldn’t be an afterthought. You need to know who is using GenAI, for what purpose, and the value it delivers. Expect your metrics to evolve as deployments mature, and don’t hesitate to drop irrelevant ones in favor of those that show actual benefits.

Partner closely with your finance team to build strong cost-benefit processes. This should include:

• A set benefits realization window (e.g., 12 months) for comparing investments
• Dedicated finance roles to validate and approve benefit calculations
• A transparent reporting process for tracking and communicating realized value

4. Ensure Adoption, Governance and Long-Term Value

Let’s be honest. GenAI won’t deliver value on its own. You need people, skills, and governance to make it stick. That means investing in training, AI literacy, and change management, so employees adopt GenAI and use it effectively. Redirect productivity gains by setting new performance goals that encourage employees to channel freed-up time into higher-value work.

Upskill or hire resources who can assess business value impacts, implement valid measurement systems, and run analyses to prove benefits with confidence.

Finally, build a strong AI governance framework to accelerate innovation while managing risk. As per a Gartner study, current patterns suggest that through 2027, 60% of GenAI projects will be abandoned after proof of concept (POC) due to inadequate AI-ready data, ineffective governance and risk controls, escalating costs, or unclear business value.

This means you need to sit down and draft acceptable use guidelines with legal teams to address IP and confidentiality risks tied to public LLMs. Ensure a strong data governance structure. Good governance doesn’t slow you down; it gives you the confidence to innovate safely and at scale.

Your Next Move

Your competitors are already moving from random experimentation to systematic GenAI investment strategies as you read this piece. Start with defensive use cases that protect your competitive position while building organizational capabilities. Then extend into applications that amplify your existing strengths and create new competitive advantages. Finally, when you’re ready, invest in transformative applications that could fundamentally change your business model or industry.

Consulting with a GenAI expert is another way to initiate the process. With a proven framework and expert-led team to guide you along the way, balancing GenAI investments with business value becomes much easier.