How to identify the best AI use cases in-house

Introduction

The key question when integrating AI into your business is, "Where to start?" or "What to focus on?" This applies whether you intend to adopt a third-party solution or implement your own GenAI, Advanced Analytics, Big Data, or any other type of solution.

This article assumes you already believe that targeted AI adoption will benefit your organization. We'll focus on how to identify and prioritize the right opportunities.

The AI Transformation Journey

Before diving into specific use cases, it's very important to get an understanding of how AI is used in practice. Without it, it's often difficult to spot valuable opportunities for AI. It's important to understand what kinds of AI applications exist and how other organizations apply them. You can learn from desk research, relevant educational courses, or by having someone provide a workshop on the topic.

When you got a couple of ideas of how AI technologies could benefit you, next step would be to prioritize and plan.

An AI transformation journey usually includes several key phases:

1. Strategy and Vision Setting - Establish a shared understanding of AI's role in the future of your organization, align stakeholders on vision and goals, and develop clear success metrics with the leadership buy-in.
2. Target Domains and Specific Use Cases Identification and Prioritization (that’s our focus in this article!) - Discover potential opportunities systematically, prioritize them, and create detailed plans for selected use cases.
3. Implementation and Scaling - Mobilize resources, plan and execute pilots, and gradually scale up AI initiatives across the organization.
4. Value Capture and Measurement - Measure the value of AI initiatives continuously (comparing metrics against a baseline), and refine your approach based on performance data.

Systematic Approach to Use Case Identification

A. Domain Prioritization

The first step in bringing AI into your organization is identifying and prioritizing the most promising domains for further exploration and development. A domain - is a part of a business that can be meaningfully analyzed and improved on its own, they can be overlapping and of different scale (examples of business domains for a grocery retail chain are: customer support, inventory procurement, pricing, physical stores maintenance, and others). This domain-level analysis helps focus your resources where they'll create the most impact.

You can approach this analysis using whatever framework best fits your organization's context. Some effective approaches to domain identification include mapping:

• The overall value chain of the business

• Customer/user journey

• Organizational processes

• Production process

For each domain, estimate two key factors:

1. Value: The potential upper limit of value creation
2. Feasibility: A general assessment of implementation feasibility

This initial prioritization requires only high-level estimates, which can be based on internal insights, hunches, or external evidence. While these estimates will be rough at this stage, they help identify the most promising areas to focus your detailed analysis and use case development efforts.

B. Use Case Discovery and Validation

Once priority domains are identified, the next step is a systematic review of them for potential use cases. Each potential use case must be validated in terms of both business value and technical feasibility.

This process includes three steps: identification, prioritization, and validation.

There are multiple techniques and research directions that can be used to support each step:

1. Ideation and discovery
   • Organize brainstorming sessions with domain and technical experts
   • Conduct stakeholder interviews and workshops
   • Analyze existing pain points and opportunities
   • Review industry best practices and study competitor solutions
2. Technical and business validation - Initial prioritization (use a value vs. feasibility matrix)
   • Define and estimate specific success metrics (e.g., net profit, customer net promotor score (NPS), throughput)
   • Assess initial technological feasibility (get feedback from technical experts)
   • Gather stakeholder feedback and refine vision
   • Estimate high-level timeline
   • Identify enablers and potential roadblocks (data availability and quality, skills, infrastructure)
3. Business case development - Final prioritization and validation
   • Define clear success metrics (you will need to measure them regularly)
   • Quantify benefits in detail
   • Evaluate required resources and investments
   • Create implementation timelines and roadmaps
   • Assess risks and plan mitigation strategies

Look for groups of AI use cases that share resources or work better together. Evaluate them as a group as well as individually. This can help uncover great opportunities (that otherwise might not make sense) since the substantial upfront investments in infrastructure and development can often be shared across multiple use cases.

At the bare minimum, the use case discovery and validation process consists of brainstorming, ball-park value and feasibility prioritization, and back-of-the-envelope sizing of costs and benefits.

Key success factors in this phase are:

• Involve both business and technical stakeholders early

• Consider both immediate benefits and strategic value

• Review and update business cases regularly as you learn more

• Remain open to new use cases discovered during the validation process

• Give more weight in prioritization to potential value if feasibility is uncertain

C. Evolution of the Process with Time

The use case identification process evolves significantly as organizations progress in their AI journey:

1. AI initiatives shift from being centrally managed to being ongoing and proactively identified and developed by individual business units.
2. Use cases are evaluated not just on individual merit, but on how they fit into and enhance the broader AI portfolio, creating compound value through infrastructure reuse and synergies with existing capabilities.

Next steps

By the end of this process, you should have a clear understanding of the next steps for the selected use cases that are validated as worth pursuing.

You then need to sequence the development of the selected use cases in time. Start with Proof of Concepts that are likely to succeed, generate significant value quickly, and are easy to scale.

However, identification is just the beginning – success requires strong execution and change management capabilities.

Want to continue the conversation?

Contact NorthLawn to explore how NorthLawn's AI Practice can support your organization's goals.