Build Your AI Business Case

Approach AI with your eyes wide open.

Analyst perspective

Be diligent in understanding all sources of AI cost and risk to realize full business benefit.

Your organization has been caught up in the AI wave and wants to implement it. What do you do?

AI is one of the hottest topics to hit the IT world, and many organizations are exploring the myriad ways to leverage it safely and effectively for business benefit. Identifying a great use case is the first step. The next is quantifying the benefits, risks, and costs of AI in a business case using as clear financial terms as possible.

This blueprint seeks to itemize all the financial factors and impacts to consider when exploring an AI initiative. With this information, your organization can enter the world of AI with its eyes wide open, confident it will gain its targeted return on investment.

Jennifer Perrier

Principal Research Director, CIO Practice
Info-Tech Research Group

Executive summary

Info-Tech Insight

AI is an organic technology designed to consume information and resources so it can grow and improve over time. Treat AI costs as organic too.

EXECUTIVE BRIEF

Your challenge

AI is one of the hottest topics in IT history. Understandably, most organizations want to reap its benefits as soon as possible.

• The broad array of potential uses for AI makes knowing where to start difficult. Requirements, available and appropriate solutions, impact on existing systems, and a treacherous risk landscape mean there’s no one-size-fits-all business case to be made for AI adoption.
• AI is a different kind of technology than what many organizations are used to. AI is organic and has a lot of moving parts, each of which require care and feeding, complicating its set up and management.
• The costs and benefits of AI are very much “it depends” propositions. Most organizations don’t realize the true costs of implementing and sustaining AI, which means total cost of ownership (TCO) calculations may have serious gaps. Nailing down the benefits that make the costs worthwhile is almost as challenging, especially from IT’s perspective, as the AI implementation is probably not for its own benefit.

Common obstacles

The AI space is rife with unknowns, which can make successful planning and rollout difficult to achieve.

• Overexcitement about AI’s potential can lead organizations to underestimate the work and costs involved. Much of AI’s power is hidden behind the scenes – a lot goes into creating any truly elegant, algorithm-based solution. This fact makes AI something of a black box. The excitement of participating in the magic can easily overshadow due diligence, planning, and transparency.
• Organizations may not be ready to adopt and operate multicomponent, computing-intensive AI solutions. AI solutions require data assessment and clean-up at the least and an overhaul of multiple IT systems, business processes, and governance mechanisms at the most. Any business case for AI must factor in these prerequisite performance and governance costs.
• An immature AI market is further clouded by new entrants and shifting price models. Product offerings, prices, and capabilities will shift and evolve quickly in the near term, bringing the usual challenges, uncertainties, and costs faced by all early adopters of an emerging technology. The business case you make today and solution you select tomorrow may be out of date by the time they are actually implemented.

“…nearly 80% of…AI projects typically don’t scale beyond a PoC or lab environment.”

Source: CompTIA

Info-Tech’s approach

Develop your AI business case one step at a time.

The Info-Tech difference:

1. Define your AI use case

   Lay out your target use case and proposed solution in detail to give your AI a clear purpose.

2. Identify and quantify target benefits, risks, and mitigations

   Identify and quantify your benefits, risks, and risk mitigations before identifying a single AI cost.

3. Identify and quantify costs to acquire, implement, and maintain

   Break your AI project and post-implementation operating costs into phases to bring maximum transparency to your total cost of AI ownership.

4. Finalize and present your AI business case

   Put your business case in a presentation-ready format that’s easy for your approvers to understand, support, and adopt.

Insight summary

Overarching Insight

AI is an organic technology designed to consume information and resources so it can grow and improve over time. Treat AI costs as organic too.

Insight 1

Have a clear understanding of your intended use case for AI – it sets the boundaries for your business case and focus for your statements of benefit.

Insight 2

Lay out your proposed AI initiative’s business benefits before tackling any project or implementation costing efforts.

Insight 3

Fully explore the project and operational risks surrounding AI, including the risks that can derail the project as well as those risks caused by the AI itself.

Insight 4

Break AI project costs into discrete phases to maximize transparency of the true sources of AI cost. Know that the biggest cost is probably not the AI itself.

Insight 5

Your final business case should showcase the inherent risks of AI, not soften them, so your organization can adopt AI with its eyes wide open and reap its benefits.

Blueprint deliverables

Each step of this blueprint is accompanied by supporting deliverables to help you accomplish your goals:

Sample AI Use Case

A complete example of how to effectively document the use case for your proposed AI solution.

Sample AI Prototyping Workbook

A complete example of a proposed AI solution that meets the needs of the target use case, including documentation of a user testing approach.

Sample AI Business Case Analysis

A complete example of the project and ongoing post-implementation costs of a proposed AI solution along with quantification of benefits and returns.

Key deliverable:

Template and Sample AI Business Case Executive Presentation

A fully editable PowerPoint template for laying out your final AI business case, populated with an example of a complete AI-solution proposal.

Measure the value of this blueprint

Set the following goals when developing and presenting your AI business case to help you measure your success every step of the way.

1. The AI business case is approved with only minor amendments.
2. Projected costs at each phase are within 5% of actual costs.
3. All identified risks are effectively mitigated with no unplanned expenses incurred.
4. Projected benefits are fully realized on schedule.

Transparency is especially critical for AI implementation success. It starts with a comprehensive, detailed, and bold business case that can stand up to rigorous scrutiny.

Document your AI cost items, benefits, and risks in your working copy of Info-Tech’s Comprehensive Business Case Analysis Tool. This is the number one tool you can use to help you define and track your AI project success metrics.

Info-Tech offers various levels of support to best suit your needs

DIY Toolkit

"Our team has already made this critical project a priority, and we have the time and capability, but some guidance along the way would be helpful."

Guided Implementation

"Our team knows that we need to fix a process, but we need assistance to determine where to focus. Some check-ins along the way would help keep us on track."

Workshop

"We need to hit the ground running and get this project kicked off immediately. Our team has the ability to take this over once we get a framework and strategy in place."

Consulting

"Our team does not have the time or the knowledge to take this project on. We need assistance through the entirety of this project."

Diagnostics and consistent frameworks are used throughout all four options.

DEFINE YOUR AI USE CASE

Understand the relationship between your use case and your business case

A use case is a specific business process or challenge to which technologies are applied in a targeted fashion to achieve a meaningful and measurable outcome.

Clearly defining the intended use case for your AI is the first step in building a business case for it. The use case:

• Defines the inclusions, exclusions, scope, and boundaries of the AI initiative you’re proposing.
• Outlines the concrete tasks to be completed by the AI, thereby informing process requirements, goals, and objectives.
• Describes what a successful outcome looks like, which in turn will dictate cost categories and amounts required to realize the use case’s benefits in the real world.

Define the use case to give your proposed AI a purpose. You need to be able to communicate this purpose not only to stakeholders but also to the AI itself.

How many use cases should you define?

• You can create and present multiple use cases if desired, including potential future opportunities or scenarios that you’ll act on if the initial use case realizes its intended benefits.
• Think of your AI as a long-term, multiphase project that will only move on to the next phase if the outcomes of the previous phase meet a series of preestablished criteria.
• However, it’s strongly recommended that you focus on a single use case to start. Given uncertainties, you shouldn’t get too far ahead of yourself. Aim for something that:
  • Will be implemented as a pilot project first.
  • Is relatively easy to describe and explain.
  • Solves a known or existing business problem.

Explore the types of AI use cases and projects you can tackle

AI initiatives can be generally classified into four types, though there is some overlap.

Workflow Automation

The AI automates existing repetitive or low-skill manual tasks.

The purpose of AI in workflow automation is to gain speed and efficiency while reducing errors and costs, usually through redeployment or elimination of staff.

Example: Robotic process automation (RPA)

Perception

The AI replaces or augments human perception such as sight, hearing, and touch, especially in situations that are ambiguous or complex.

This type of AI is often applied to customer service or security scenarios. While it can reduce a lot of “noise” and dramatically expand image/audio searches, ensuring the AI’s accuracy and eliminating its bias requires intensive training.

Examples: Natural language processing, image classification, and recognition

Generative

The AI creates new content based on examining the patterns and structure of similar content.

Generative AI’s main purpose is creative. However, it can compromise a content creator’s credibility and reputation and presents a series of copyright, ownership, and privacy issues. The big concern is the ability of bad actors to use it to create false or fake information.

Examples: Writing, art, design, coding, advertising, and entertainment

Analysis & Knowledge Engineering

The AI analyzes information to identify patterns and develop new models and insights that describe or are hidden within it.

This type of AI holds great promise in scientific and academic research, product design and development, risk management, and planning.

Examples: Scenario analysis, predictive/probability modeling, and decision support

Like use cases, there are four general business goals for AI implementation

All four types of AI have a role to play in advancing business goals, but some are more compelling than others. In general, AI that enables analysis and knowledge engineering offers the highest potential rewards across the range of goals.

Improve Operations

• Workflow Automation
• Generative
• Perception
• Analysis & Knowledge Engineering

Reduce Risk

• Analysis & Knowledge Engineering
• Workflow Automation
• Perception
• Generative

Support Revenues

• Analysis & Knowledge Engineering
• Workflow Automation
• Perception
• Generative

Drive Innovation

• Analysis & Knowledge Engineering
• Generative
• Perception
• Workflow Automation

Workflow automation is the current leading AI use case

The more embedded an AI capability and adopted an AI use case, the more proven and defensible they probably are. This makes workflow automation the safest bet.

It’s no surprise that robotic process automation (RPA) is the most embedded AI capability to date. It’s followed by computer vision, which is the analysis of visual inputs like images and videos in support of decision-making.

Natural language text processing and virtual agents are a close third and fourth and are likely the AI capabilities most used by regular consumers in their day-to-day lives.

Not surprisingly, service operations optimization is the most commonly-adopted AI use case. Customer segmentation and customer service analytics are two other related use cases that have gained traction.

What’s interesting is the adoption of AI by businesses to create and enhance new and existing products. Not only are organizations leveraging AI for their own internal processes but they are promoting and disseminating it in their offerings.

Source: Stanford University, 2023

At the heart of most AI use cases is a redefinition of business processes

Some business processes will be reinvented by the AI, and some will need to be reinvented to support the AI. Here are some AI-affected processes across functions.

Finance:

• Accounts receivable/payable, expenses, reconciliation, and closure of financial ledgers
• Audit and compliance
• Fraud detection
• Automatic requisitioning of services/supplies from a third-party preferred supplier
• Provisioning of costs within the budget
• New project planning

HR:

• Talent sourcing, selection, and recruitment

Legal:

• Legal discovery
• Contract reviews

IT:

• Service desk
• Security
• Automated operations

Product design:

• Generative design
• Content creation

Operations:

• Intelligent routing and processing of customer requests
• Workforce scheduling
• Predictive maintenance
• Inventory management
• Demand forecasting

See Info-Tech’s Get Started With Artificial Intelligence blueprint for more examples of applied and narrow AI in business.

Start outlining your AI use case by determining the problem you’re trying to solve

Ground AI dreams in reality by identifying a concrete objective.

1. Identify the business problem you’re trying to solve or opportunity you’d like to exploit. Think about the types of AI available, common AI goals, and specific business processes and ask yourself the following questions.
   • What top business opportunity do you wish to see realized with AI?
   • What top risk will AI help you alleviate?
   • What top annoyance can you relieve with AI?
   • What kinds of new insights will you be able to gain with AI?
2. What are the challenges of carrying out this activity or process in its current state?
3. What would resolving these problems and challenges using AI look like?
4. What other business activities and processes could be improved if you resolve these problems and challenges?

Download a copy of Info-Tech’s Use Case Template to outline the business process you’re tackling as well as any special requirements, assumptions, and issues. Also see this Sample AI Use Case for specific guidance.

Next, outline your proposed solution to the problem at a high level

This is where you introduce AI to the real-life execution of a business process.

Ask these questions:

1. What solution are you proposing at a high level?
2. Who will use the solution and/or be affected by it?
3. What are the steps involved in using the solution?
4. What are the intended outcomes of the solution?
5. What is net new about the solution or newly enabled by it?
6. What will the solution reduce or eliminate?

Develop a character persona when defining your use case. A character persona is a fictional person that exemplifies a typical user of the proposed solution. This helps in two ways:

• You’re better able to capture the key activities, steps, and overall experience from the user’s point of view. This helps ensure nothing’s overlooked and communicates that you understand what the user experience could be like.
• If the persona is realistic and relatable, your target stakeholders will be better able to visualize your proposed solution, facilitating buy-in and support.

The number one obstacle you should expect to encounter is outright resistance on the part of prospective users, especially if the proposed solution is coming from IT and not from the business itself. Simply put, some people are going to hate it even if it promises to make their jobs better in the long run. Plan to put a lot of effort into stakeholder care and run a rigorous proof of concept before going all in.

Note your core assumptions and constraints

No use case is without complications. Identify factors affecting the current process that you’ll need to address for the proposed solution to succeed.

Ask these questions:

1. What compliance, regulatory, or policy concerns do you need to consider in the solution?
2. What existing business strategies or goals are supported by the effort and need to be considered and/or cited?
3. What performance measures are impacted and/or must be tracked throughout the use case?
4. What are the applications, systems, and integration points at each step in the use case? What already exists, what will require modification, and what will need to be procured?
5. What data elements are either created, used, or transformed at each step in the use case? What will need to be done to them to make them usable by the AI?

It’s imperative to identify both assumptions and constraints as these clarify what you can and cannot do, effectively setting the boundaries of your proposed solution. When pitching your solution, communicate your assumptions and constraints to let your stakeholders know that you’ve considered all the angles and provide the logical foundation for the decision you’ve made.