To experience what AI can do for your business, let’s step away from online conversations for a moment. While AI is powerful, it can quickly turn a good thing bad or have no impact at all without a well-defined strategy.
As cliché as it sounds, it all starts from within. Your unique challenges and business context hold the key to unlocking one-of-a-kind AI opportunities.
Why is Finding the Right AI Use Cases so Difficult?
The short answer is that it's meant to be, at least at first.
O'Reilly's annual survey on AI adoption reveals that identifying appropriate use cases has been a significant bottleneck in AI adoption for the past few years.
Why is this important? These statistics shed light on a couple of aspects:
Firstly, awareness exists, which usually leads to positive change. Businesses that acknowledge finding the right AI use case as a challenge are more likely to delve deeper into the underlying issues, seek external help, and explore their options.
The fact that finding relevant business use cases remains a bottleneck year after year highlights the scarcity of resources to address it. The lack of frameworks or experience makes identifying the right AI use cases difficult for most companies.
Choosing AI use cases shouldn't be a random act of imitation or choice. To pinpoint areas where AI can most benefit your business, you must have well-defined criteria and a shared understanding of how AI can contribute to your specific context.
And there are two main aspects to choosing the right AI use cases: identification and prioritization.
We've compiled a few guiding principles to help you reduce uncertainty and the risk associated with finding the wrong AI use cases. These insights will help you look into the impactful ones adapted to your organization's needs.
Identifying AI Use Cases: Finding the Sweet Spot
So if it’s not a matter of imitation, and you also don’t have a blueprint to start with, identifying the right use case might seem like looking for a needle in a haystack. It shouldn't have to feel this way. These are some of the steps we typically take with our clients during this process. You might find them helpful.
1. Determine Strategic Importance
Begin by examining your company's overarching strategic priorities. What are your goals for the next 3-5 years? Depending on whether your strategy focuses on addressing current challenges (e.g., reducing production costs) or exploring new opportunities (e.g., entering new markets), you may discover different use cases.
Identify the business areas with the highest ROI potential, and see if they align with your company's overall strategy. This approach encourages stakeholder buy-in, optimizes resources, accelerates adoption, and ensures that innovation doesn't happen in silos.
Let’s take a look at an example of a situation many of our clients face when identifying strategic business areas.
When we talk about the quality control of mechanical components, the few experts who can assess the quality of the components often use their hands and ears to listen to and feel (haptic control) the components. By doing so, they determine whether or not the components are up to par, and if there’s an issue, identify the potential cause.
Imagine that your company is growing and you need to open a new manufacturing site. And since you’re relying on a manual quality control process, here are a few issues you may encounter:
(1) With the new location and higher salaries, it would be too costly to continue manually inspecting components at the new production facility.
(2) It may be that your seasoned experts who have been at your company for a long time, don't want to transfer to a new manufacturing site to train the new employees.
An otherwise smooth quality control process that nobody notices, unexpectedly turns into an obstacle to the long-term growth of the company, becoming a strategic concern.
2. Pinpoint Bottlenecks
Compared to identifying key strategic use cases, pinpointing bottlenecks takes a bottom-up approach. It’s much more focused on operations and the nitty-gritty problems of your everyday processes. It involves taking a look at your organization's challenges.
What obstacles are preventing you from reaching your operational goals? You might want to focus on slow, inefficient, or error-prone processes that involve highly-repetitive tasks or customer interactions. Alternatively, you could target operations ready for the next stage of innovation. There’s a good chance you have already compiled a list.
Once you have identified these challenges, consider how AI might help resolve them. Successful AI implementation doesn't depend only on choosing the right use cases, but also on understanding the AI/ML technology's capabilities and limitations.
When selecting an AI use case, outline your expectations and how you believe AI can address the identified problems – an approach that will help you set realistic goals, allocate resources efficiently, and evaluate the project's success once implemented.
The typical use cases here are operational, focusing on top- and bottom-line growth through efficiency gains, increased customer engagement, risk mitigation, and regulatory compliance.
Here's what we mean:
Let’s say you’re facing irregular and unplanned stoppages in your production line that negatively impact your OEE. So you would like to:
1) Find out what causes the stoppages, and
2) Predict when these stoppages might occur with a forecasting horizon that allows your shop floor teams to react.
In both cases, root-cause analysis, advanced analytics, and AI algorithms can be applied to identify the underlying issues and develop effective strategies.
Pharma companies may find the following situation familiar:
Medical Information (MI) / Medical Affairs (MA) teams spend increasing amounts of time and effort scientifically reviewing promotional material. And there are a lot of reasons contributing to this situation, including greater regulatory oversight, a larger treatment portfolio, or increased FDA enforcement.
Artificial Intelligence (e.g., Natural Language Models) might greatly simplify the review of promotional materials, freeing your MI/MA team's time for more value-added tasks.
3. Understand the Industry's Direction
It's easy to be drawn to success stories, especially when they occur in your market. We've all been guilty of looking at a competitor's achievements and thinking, "If it worked for them, surely it must work for us too." However, there's more to successful AI projects than meets the eye. Data quality, team expertise, operational structure, and change management processes are just a few factors that determine the success of an AI project, and no two companies are exactly alike.
Besides, it's sometimes difficult to distinguish between marketing stunts and an effective ROI.
While this is true, staying informed about industry trends and successful AI implementations within your sector can provide valuable inspiration for identifying untapped opportunities and potential risks. Attend industry conferences, listen to other experts in the sector, and follow your competitors – all by keeping an open mind.
4. Evaluate Your AI Readiness
The ideal use case should align with your current context and capabilities while at the same time nurturing your AI readiness, taking you to the point where AI is completely embedded in your business's operating system.
So rather than focusing on your competitors, be driven by your current bottlenecks and strategic goals so you can identify use cases that will most effectively build your AI maturity.
Listed below are just a few project ideas that we’ve seen help companies become more AI-ready:
Coach teams on how to use ChatGPT in their daily work - advancing their abilities
Consider re-designing or designing new components and adding them to your data infrastructure to facilitate the collection and aggregation of data for downstream AI tasks (as opposed to an often encountered IT perspective, where data collection is primarily about storing data rather than generating insights or even performing prescriptive analytics). These types of initiatives will build your technology maturity.
Check out this post to learn the differences between a data lakehouse, a data lake, and a data warehouse.
Striking this balance requires a solid understanding of AI's potential and limitations, as well as the role your company's data plays in successfully implementing an AI solution. To assess your current state of AI adoption, consider where you stand on the AI maturity curve and prepare for the next stage.
Progressing to the next stage of AI maturity may vary for each organization. To determine whether your use cases and AI maturity align, consider the following questions:
Will you need to make any data infrastructure updates?
Do you have a data engineering/data science team? Can you build or enhance one? Would you consider bringing in external help?
Do you have a change management process in place?
What are the ethical and regulatory implications? Are you prepared to address them upfront?
5. Assess Data Readiness
Successfully scaling an AI project relies heavily on a company's data readiness. Whether you're exploring a potential use case in Customer Service, Manufacturing, or R&D – it needs to be centered around data; it must be a data problem. AI/ML models depend on high-quality, well-structured data. Assessing the availability and quality of the data required for a specific use case from the start can save you time (e.g., avoiding delays) and resources (e.g., the cost of data collection or cleansing efforts) later.
Many use cases in this dimension revolve around the statement: “We don't know what we don't know”. And that means that the task of analyzing, often, unstructured data, is to identify patterns you weren’t aware of.
A pharma company's medical staff would collect data through conferences or phone calls with other medical professionals, or sales reps would collect information through different touchpoints with healthcare professionals. As you’re applying this case to your business, you’re thinking: “I could start analyzing this unstructured data using word clouds”. But word clouds are a bit 2000, aren’t they? Now we can use advanced technologies, such as Large-Language Models (LLMs) to uncover hidden patterns in this data and validate business hypotheses.
In other words, when you let the data speak for itself and remove human bias, you can start uncovering your data's true predictive power.
The right AI use case lives at the intersection of multiple dimensions, which may or may not be on your radar depending on your role within the company. Finding them is most effective when approached through cross-functional collaboration. As a result, the chosen use cases will be comprehensive, cater to different perspectives, and maximize success across the organization while involving all stakeholders in the change management process.
Prioritizing AI Use Cases: Ticking All the Boxes
Setting clear criteria for what makes a good AI use case can help you prioritize and select the most suitable options. However, the prioritization process is not universal and may differ for companies at various stages of AI adoption.
Whether you're looking for short-term wins or long-term strategic advantages, the following criteria can guide your AI use cases evaluation:
Ensuring clear strategic alignment means the AI project directly contributes to your business objectives, adding genuine value to your operations.
Consider the potential return on investment (ROI) that each AI use case can offer. The potential savings or gains from the AI solution should justify the investment made. While some use cases may appear attractive or promise quick wins, ask yourself: will saving a few minutes in that process truly make a difference?
AI use cases should have the potential for future development. It's not just about meeting immediate needs; the chosen use case should be capable of growing and evolving alongside your business. As your data increases or your needs change, the AI solution should be able to scale accordingly. A scalable use case is one that other teams can also benefit from.
2. Ease of Implementation
Assess the technical requirements, data availability, and quality needed for the AI use case. Not all businesses have the same technical capabilities or access to high-quality data. Make sure your company has the necessary resources or can feasibly obtain them before committing to a specific AI use case.
Time & Effort
Take into account the complexity of the use case and whether your business can execute it. It may involve factors such as project duration, the need for technical know-how or domain expertise, or potential disruptions to existing operations.
3. Stakeholders Buy-in
Do the use cases have strong support from leadership, IT, and subject matter experts (SMEs)? Early involvement and clear communication about the benefits of the AI project can help secure stakeholder buy-in, commit to its success, and facilitate a smoother adoption process.
Consider the impact of the AI use case on your employees. Will it help them work more efficiently? Will it require them to acquire new skills? Understanding these factors can help you address potential resistance and ensure employees see the AI project as a beneficial tool rather than a threat.
By following the strategies outlined in this guideline, you can balance the quick-win opportunities with complex, high-impact AI use cases and build a roadmap that gradually advances your capabilities, aligns with your strategic goals, and consistently adds value to your operations.
Subscribe to our newsletter
You’ll receive insights, strategies, and best practices that help you succeed in adopting and implementing AI & Data. Only what matters. Once a month.