Assessing the Business Potential of Generative AI — Generative AI Value Framework
By Marco Cigaina and Itzhak Shoshan — SAP Intelligent Enterprise Institute
Introduction
In our previous article, we discussed a business-centric approach to identifying unique applications of Generative AI within enterprise operations. In this article, we will provide insights on how to effectively assess the business value of identified use cases.
In fact, lately Generative AI has emerged as a transformative technology, offering immense value across a variety of business functions and industries. However, evaluating its full impact requires a structured approach that captures this technology’s multifaceted contributions. The suggested Generative AI Value Framework is designed to offer a comprehensive, actionable tool for organizations to assess the value proposition of Generative AI.
Value Framework
The value framework for Generative AI applications is an essential tool for enterprises aiming to outline and assess the value of Generative AI solutions. It details an integrative construct that aligns AI’s practical applications with overarching business goals. The framework emerges from an understanding that AI’s true value is not solely captured by measurable impact on business operations but expands also to decision-making processes and market competitiveness.
Here is an overview of the framework:
At the core of it lies the “Value Architecture” phase, which is at the beginning of the value assessment process and is specific to Generative AI. The remaining phases (including defining a baseline measurement, calculating implementation costs, estimating benefits, calculating the ROI, and conducting an ongoing value tracking) follow a standard value assessment approach, as these steps are not specific to Generative AI, but rather general.
In this article we shall dissect the “Value Architecture,” which is designed as a consistent set of interrelated components, each contributing to the collective objective of devising a comprehensive assessment of the value derived from Generative AI solutions.
The Value Architecture Building Blocks
Before diving into the specifics of the framework, it is crucial to understand its three main components: Value Categories, Value Drivers, and Value Patterns. Here is a short description of these components, which will discuss later in more detail:
- Value Categories serve as the cornerstones for assessment, encapsulating broad areas such as revenue growth and cost savings where Generative AI can offer tangible benefits.
- Value Patterns categorize the impact of AI on the business, breaking it down into sub-components that offer actionable insights.
- Value Drivers operationalize these categories by identifying real-world applications in various industries, offering a detailed map of where and how value is generated.
Together, these three elements form a holistic framework for evaluating the multifaceted value that Generative AI brings to an organization.
Value Categories: The Cornerstones
We identify five categories of value as foundational for our framework: revenue growth, cost saving, sustainability, time savings and risk reduction. Each category is accompanied by relevant KPIs that serve as quantifiable metrics for businesses to measure the impact effectively. Here are some examples of those KPIs:
- Revenue Growth KPIs: Revenue per employee, Customer Lifetime Value (CLV), Conversion Rate
- Cost Savings KPIs: Cost per transaction, Operational Efficiency Ratio
- Sustainability KPIs: Greenhouse emissions, Recycling ratios, Raw Material to Waste Ratio
- Time Savings KPIs: Average Handling Time, Employee Productivity Rate
- Risk Reduction KPIs: Compliance Rate, Security Incident Frequency.
Value Patterns: The Archetypes of Value
The framework’s core lies in Value Patterns that abstract from specific metrics of value and categorize AI business impacts.
Value patterns identify and describe specific ways in which generative AI contributes to business performance. Each pattern is characterized by its ability to be replicated and applied to different scenarios within the company, allowing businesses to anticipate the potential benefits and strategize their investments in AI technologies accordingly. They serve as guidance for decision-making and action-taking that can be recognized and implemented across multiple business contexts to achieve tangible and measurable value.
The Value Patterns are mapped into a view of end-to-end processes that is taken from SAP Signavio Explorer. This view organizes end-to-end processes in four activities: develop product and services (in short, the “Product & Services process area”), generate demand (the “Customer” process area), fulfill demand (in short, “Supply”), and plan and manage the enterprise (in short, the “Corporate” process area).
Each pattern is accompanied by specific KPIs, examples, and explanations, making the framework actionable for businesses. The following tables recap the key Generative AI Value Patterns.
Operations Value Patterns
Products and Services Value Patterns
Customer (Generate Demand) Value Patterns
Supply (Fulfil Demand) Value Patterns
Corporate Value Patterns
The preceding Value Patterns serve as an initial framework for assessing the multifaceted impact of Generative AI on business value. It is important to note that the landscape of artificial intelligence is in a constant state of evolution, and as such, we anticipate that additional value patterns will emerge. These could encompass new dimensions of business value not yet fully realized or understood. Therefore, this framework is designed to be dynamic and adaptable, accommodating for future inclusions as Generative AI continues to mature and proliferate across industries.
The Logic of Value Patterns
When evaluating the business value of Generative AI use-cases, it is essential to categorize the types of impacts they create. Generative AI can significantly impact business operations and since the affected business operations can be seen as sets of tasks, the value of a Generative AI use-case on business operations can be estimated by calculating the gain per task, subtracting the cost of validation per task, and then multiplying by the task volume. However, it is crucial to differentiate between various scenarios.
Firstly, we must determine whether we are discussing an existing capability or a new one. Existing capabilities can be further divided into common capabilities (tasks that many employees can perform) and scarce capabilities (tasks that few employees can perform, requiring external assistance).
In the case of a common capability, we can further differentiate between a scenario where an AI application enables an employee to complete a known task faster, and one where AI improves the quality of the task outcome.
We categorize “efficiency gain” as a value pattern that encapsulates all value drivers where cost-saving is achieved through accelerated task completion. For instance, Generative AI can summarize a text much quicker, so in that case the estimated value can be calculated by multiplying the time saved per task by the cost of time, subtracting the validation cost per task, and multiplying the result by the task volume.
“Quality improvement” is another value pattern that includes all drivers that derive their value from enhancing the output quality of a task: for example, the increased accuracy of a prediction, classification, or forecast. The estimated value can be calculated by subtracting the validation cost per task from the added value of the improved quality, and then multiplying the result by the number of tasks.
“Skill enhancement” is a value pattern associated with using AI to augment employee capabilities with new skills. For instance, AI can eliminate language barriers, enabling all employees to communicate in any language. The total cost of either outsourcing a service or acquiring the necessary skills, multiplied by the task volume, determines the value.
However, if a new operational task is made possible solely due to AI, the entire value of that task should be considered. For instance, if Generative AI enables real-time reaction to social media at scale, which was previously impossible, the whole value of this task should be acknowledged.
The following diagram recaps the logic we just described.
Another type of impact pertains to the use of AI in a company’s product or service portfolio. If AI is used to enhance a product (like a better digital assistant in cars), the added value of the product, derived from the premium price and increased sales due to the new feature, should be considered. In some instances, AI could enable an entirely new solution or business model. For example, an AI-powered robot-taxi service. In such cases, the complete profit from the new solution or business model should be included in the value assessment.
The impact on customer experience and satisfaction should also be emphasized. For instance, a customer service chatbot powered by Generative AI can help resolve product issues, improving the net promoter score, which influences customer churn, thereby affecting costs and revenue. AI also contributes to increase market responsiveness and agility by analyzing data in real-time and adjusting specific parameters of business (e.g., pricing), thus contributing to accelerating time to market, ultimately gaining market share.
When we consider the process area of demand-fulfillment, we need to consider procurement, manufacturing and distribution for product companies and service delivery for service companies. In both cases we should consider the impact of AI on optimizing resources and sourcing.
Finally, we need to focus on the impact on the corporate management area. In fact, Generative AI can improve strategic decision making, for example with better simulations of business opportunities. Furthermore, in this era of triple bottom lines, it is also crucial to consider the impact of generative AI on sustainable development goals: Generative AI can increase compliance, generate sections of the sustainability report, and reduce emissions by designing new systems. Sustainability should be systematically considered across all Value Patterns. Generative AI can also improve risk management and business continuity. Finally, we need to consider corporate management of resources. Looking at human resources, we need to consider the improvement of employee engagement as a value impact of relieving employees from repetitive tasks. Looking at material — both tangible and intangible — resources, we need to consider the impact related to unlocking value from data.
Business Transformation Context
The framework we presented serves as valuable resource for organizations aiming to harness the full potential of Generative AI in their business operations. Value creation is a key pillar of the Business Transformation Methodology, a methodological framework from SAP Business Transformation Services. Our considerations about the value of generative AI in a business context, can be leveraged within the Business Transformation Methodology to accelerate the value discovery and business case associated to a use-case based on Generative AI.