BEST PRACTICE

Identify Data Analytics & AI Use Cases

In this article, we dive into learnings and best practices that assist you in identifying strategically suitable and value-adding Data Analytics & AI Use Cases for your business. This is one of the most crucial steps in protecting against unplanned expenses and achieving a positive return on data & AI.

Read time: 9-11 min.

Topic Intro

Predicting trends, automating processes, or enhancing customer experiences. The analysis and processing of data using a variety of algorithms, particularly through Machine Learning models, is not just an investment in the future but also a crucial competitive factor in the present. For modern businesses, Data Analytics and AI Use Cases are already of great importance today in order to reduce costs, enhance core business, or create entirely new revenue streams.

However, it would also be incorrect to ignore the fact that not every Use Case idea develops into a successful operational solution. The failure rate still stands at around 80%, which should not be surprising in an experimental and high-tech environment like Data Science & ML. This rate becomes problematic, however, when companies fail to sufficiently anticipate foreseeable hurdles, leading to avoidable costs. For instance, identifying unsuitable Use Cases …

Table of Content

Chapter 1 | Why do companies struggle in identifying use cases?

Chapter 2 | What do experienced companies achieve?

Chapter 3 | A proper preparation is key

Chapter 4 | Collect new ideas

Chapter 5 | Sorting ideas and structuring the backlog

Chapter 6 | Conclusion

Chapter 1

Why do many companies struggle to identify suitable use cases?

With Data & AI, much is possible and can create significant competitive advantages. But how do you distinguish hype from reality? The identification of a practical AI use case and its alignment with strategic business objectives is crucial for the success of AI initiatives. However, companies still find this challenging. Various studies, whether from Gartner, O’Reilly, or McKinsey, prove that identifying suitable use cases is a significant hurdle in implementing AI and achieving a return on investment.

“The first project that the CEO suggested is not the right one to invest in.” — Andrew Ng during Amazon re:MARS 2019

The rapid advancement and simplified applicability of technology, as recently demonstrated by the successes of Large Language Models, make it difficult for companies to focus on specific projects. #FOMO It is also a well-known approach to start with an extremely ambitious project to demonstrate the full potential of this technology, what is possible, and how innovative the company is.

The result is that following impressive initial demos, there is often significant disappointment. Either the technical scaling is enormously challenging due to the lack of appropriate infrastructure, or the maintenance effort escalates (consider data and model drift). And more often, it’s classic hurdles in internal processes, such as a lack of acceptance or unclear responsibilities, that come into play. Thus, Data and AI Use Cases can quickly become failures; the desired success does not materialize, and investments in these technologies are called into question again.

The Reasons Why Data & AI Projects Fail:

Lack of understanding of business problems: The defined problem space is always the initial point of a Use Case. It requires extensive communication with stakeholders such as customers and employees and methods to clearly define and prioritize business problems.

Insufficient technical understanding: There can be several different analytical solutions, including AI and machine learning, that can potentially solve the same problem. So how do you choose the right one?

Unclear AI strategy: To identify the AI Use Cases that add value to your company, a common understanding of why, how, and where your company wants to use AI is needed. Where will we play; how will we win; what capabilities do we need?

Lack of internal communication: Only when departments recognize that data and AI are crucial future aspects of corporate strategy, understand the benefits of these technologies, and have clear points of contact available, will practical and value-creating application possibilities for the use of AI be recognized (Bottom-Up!).

Insufficient knowledge transfer: If models and experiences are not or hardly present, the identification of Use Cases is usually marked by uncertainty and risk. Missing templates and rough process steps with responsibilities are hurdles for drive and professionalism.

Lack of self-reflection: Companies that recognize the challenge of finding the right AI use case are more willing to address underlying problems and, if necessary, rely on external expertise.

Chapter 2

What benefits do experienced companies achieve?

Overall, identifying relevant Data Analytics & AI Use Cases is fundamentally important to fully leverage the potential of these technologies and ensure they support sustainable business success. Companies that successfully meet this challenge are clearly better equipped to thrive in an increasingly data-driven world. Specifically, identifying suitable Use Cases can yield a range of benefits for your company.

Promotion of internal knowledge & experience
Development of data culture and demystification of AI
Demonstration of innovation capability internally & externally
Adaptation to market changes
More effective use of resources
Better consideration of risks
Reduction of misinvestments or faster generation of a return on investment for Data & AI initiatives
Acceleration of the implementation of strategic business goals

Chapter 3

A proper preparation is key

The preparation sets the foundation for the entire process of identifying suitable use cases.

Assess your Data & AI Maturity Level

How can you plan for the future without knowing where you are & where you need to be? The ideal use case should fit your company’s existing data & AI capabilities and at the same time strengthen and enhance them. It’s important to know the Data & AI Maturity Level of your company. This includes evaluating aspects such as data sources, tech stack & infrastructure, governance, culture, and talent. Be aware that these key enablers are the prerequisites for the successful development of a use case. And that the right use case truly unlocks the potential of these key enablers and harmonizes them.

Assessing the ‘Data & AI Maturity’ of your company supports identifying further strengths and weaknesses. Specifically, it helps you to find use cases that seamlessly integrate into your existing infrastructure, offering added value. Or, for more challenging use cases, risks can be better assessed and preparation costs more accurately calculated.

Ask yourself for example (just to name a few):

What is the status of data availability and aggregation?
How successfully is the data governance system integrated?
Are you ready to tackle regulatory issues and compliance processes such as the GDPR or the AI Act?
What data science, machine learning or data engineering skills do we have in our company?

So, don’t focus on your competitors, but let your current bottlenecks and strategic goals guide you, so you can identify use cases that most effectively promote your Data & AI Maturity.

Need support in assessing Data & AI maturity?

If you are interested in the topic of AI Maturity Assessment, our experts and partners will be happy to help you.

Contact

Keep an eye on strategic priorities

Begin by first giving yourself, your team, and your stakeholders a uniform understanding of the overarching strategic priorities. So, take the company strategy into account, and look at the short, medium, and long-term goals and plans of the different business units.

Tip: Create a Strategy Impact List

Define appropriate strategic benefit areas (e.g., cost minimization, core business revenue growth, new business models, customer satisfaction, ESG).
Assign these to individual strategic goals.
Evaluate the ROI potential for each business unit (e.g., “high, medium, low”) to facilitate subsequent prioritization.

As a result, you should now have a list of all strategic goals, sortable and filterable by… time to goal achievement, business unit, ROI potential, and benefit area. This is a very good basis for steering further stakeholder communication and providing prepared inputs for later brainstorming sessions. But above all, this ensures that use cases have a direct link to strategic priority areas.

Understanding the operational problems of the individual teams

In contrast to strategic goal analysis, identifying bottlenecks and challenges for operational teams follows a bottom-up approach. This is more about day-to-day operations and the minor problems in daily workflows. What obstacles prevent you from achieving your operational goals? This approach will help you set realistic goals, allocate resources efficiently, and assess project success after implementation. Typical use cases here are operational, focusing on growth in top and bottom-line areas through efficiency improvements, increased customer loyalty, risk reduction, and compliance.

Tip: Know who your sponsors are
For the success of initiatives, especially in the early stages, it’s crucial to identify business stakeholders who are motivated and look forward to data & AI topics. Go out and talk to them. Develop the first use cases in their areas of responsibility.

Effective communication

One thing is certain: communication is a central factor in the process of advancing data & AI initiatives. Only through effective communication will you be able to identify and develop relevant use cases from various business areas of your company in the long term. Pay attention to the following points in this process.

Explain and make the concept of data and AI accessible. Especially regarding artificial intelligence, there are often very different ideas about the possibilities and limitations of this technology. It is extremely important for you to give your stakeholders in the company the opportunity to learn more about the technology and demystify it. This can start with several brown bag lunch sessions, where you present various concepts or general examples of machine learning at a high level. Training sessions and seminars are also important elements to create a common horizon of expectations and to be able to proactively educate oneself.

Is an AI vision defined and communicated? Having a clear vision for the application of AI is extremely important in order to continuously align efforts towards a common goal and set priorities. For example, a vision could be: “Trustworthy artificial intelligence should significantly improve the daily work of our employees.”

Chapter 4

Collect new ideas

Now it’s about understanding strategic goals, operational problems and the possibilities of AI and combining them into use case ideas

Who should be involved?

Identifying viable AI use cases is a team effort and requires interdisciplinary skills. Organize at least a half-day workshop with teams from the respective business areas. It is crucial to bring together business domain experts and AI experts, as both possess specialized knowledge in business and technology.

The business domain expert deeply understands the business and the context/systems surrounding the business processes and would be able to grasp the business value AI will bring.

An AI expert (Data Scientist or ML Engineer) is someone who has worked with AI and is deeply knowledgeable about its capabilities, technological advancements, and successful adoption of AI technology in the industry. They are also competent enough to provide guidance regarding the complexity and feasibility of the AI use case

The Use Case Workshop

The AI Use Case Workshop aims to deepen the understanding of AI and explore innovative use cases through collaborative efforts.

1) The workshop begins with a detailed introduction to AI basics, where participants are enlightened about essential concepts, techniques, and applications.

2) In the following stimuli session, real AI application examples are presented to stimulate the creative thinking of the participants.

3) Next, the business unit presents its strategic goals and operational challenges, with a special focus on processes characterized by inefficiency, error-proneness, or repetitiveness. A visual representation of these aspects during the workshop is essential.

4) Inspired by the Design Thinking method, the creative core of the workshop follows. Participants are integrated into the idea-generation and problem-solving process. Open sessions can be conducted where ideas are spontaneously introduced, or participants systematically work through previously identified problem areas.

5) Ideas are ideally systematically collected and organized in relation to the respective problem areas or strategic goals. An initial prioritization of ideas is made in the group to create a manageable list of 10-25 concepts for further elaboration.

This promotes a comprehensive understanding of AI and provides specific and actionable insights for future endeavors. For effective prioritization, the procedures should be simple and clear. A maximum of 10-12 participants is recommended to ensure a focused and productive working atmosphere.

Another example of generating ideas
Employees can submit their ideas and identified problems via a central platform. Predefined templates help to collect the most important information of a use case idea and provide helpful orientation for the user. An internal communication campaign or a competition can raise awareness and increase participation.

Documentation of initial ideas

The initial ideas must be carefully and systematically documented. This process can be a joint effort between business experts and AI specialists to ensure that all relevant aspects are considered. For the further development and understanding of an initial idea, the following information is particularly important:

Problem Definition: A clearly outlined picture of the challenge or problem to be solved. Proposed Solution: A sketched representation of the solution idea, even if it is still in the developmental stage. The Outcome Statement following the Outcome-Driven Innovation Process is always a good compact solution.
Value proposition: An initial assessment of the potential value for the organization, including allocation to predefined benefit categories such as cost reduction, revenue increase, new business models, customer satisfaction, or ESG.
Data and Resources: Detailed information about the available data and the resources needed to implement the idea.
Assignment to the organizational unit and contact persons: Clear indication of which area of the organization is affected, including the contact details of the responsible persons.

This structured approach ensures that each idea can be further elaborated and evaluated subsequently and lays the foundation for the next phase of conceptualization. Moreover, this ensures that the ideas correspond to the strategic directions of the company.

Chapter 5

Sorting ideas and structuring the backlog

The efficient organization of use case ideas is essential in order to derive the greatest possible benefit from them.

Systematically build up an idea backlog

A well-organized idea backlog for Data Analytics and AI enables teams to systematically manage innovative ideas, drive implementation, and measure successes. This not only promotes the company’s innovation strength but also the coherence and synergy between technology investments and business objectives. It also fosters transparency, thus encouraging open communication and collaboration among team members and stakeholders. Everyone can see the progress and status of the ideas, contributing to informed decisions and better alignment of initiatives with corporate goals.

What should be considered?

Create a single point of truth that is visible to a broad community of your company and can be managed by specific individuals such as Product Owners, Team Leads, or Portfolio Managers. Tools like Casebase, but also Excel spreadsheets, can assist you in this. Transparency about ideas and processes is a value that promotes bottom-up and acceptance processes within the framework of data culture.
The backlog should contain specific criteria that allow for a comprehensive statement about the idea and enable the evaluation of its technical and business suitability. In the previous chapters, criteria for strategic mapping of business goals and operational pain points were already mentioned, as well as basic information for the idea description. Use this meta-information of an idea as a criterion for your backlog.

Communication & Transparency
It´s furthermore essential to publish the process, the selected ideas, and the next steps within the company. This promotes a corporate culture of openness and innovation and enables collective participation and responsibility.

Prioritizing the idea backlog

Now that all ideas are systematically documented in one place, it is all the easier to obtain an initial assessment of these ideas from business and tech stakeholders as the basis for prioritizing your idea backlog. The prioritization serves the question of which use case idea should be pursued next, i.e. resources should be reinvested in order to develop an analytical concept in the next phase (hypotheses, data preparation, potential models, and features).

Here it makes sense to divide the evaluation criteria into the dimensions of “business value” and “technical feasibility”.

Value:

Business Impact: Does the use case result in financial, social, or ecological value?
Strategic alignment: How strongly does the data analytics or AI use case idea support the strategic goals?
Process Impact: How helpful is the use case idea for operations?
Ethical or regulatory issues: Are there any ethical or regulatory concerns?
Innovative strength: Is this idea more incremental or disruptive?

Feasibility:

Data availability: How easy is it to access the required data?
Availability of talent: Can we process this use case using our own resources or do we need external support?
Infrastructure: Do you have the infrastructure and environment to support AI work?
Algorithmic complexity: What is the complexity score for the AI solution? Is a long training time required? Will there be a multi-stage AI pipeline?
Deployment challenges: How complex is implementation into an existing product or process?

Finally, have the assessments of both dimensions visualized on a priority matrix. This will give you a clear overview of the value and feasibility of each idea and your entire backlog. And, of course, you will ultimately have an optimal basis for prioritization decisions and stakeholder communication.

The systematic identification of relevant data analytics and AI use cases is now complete. Now it’s time to plan your roadmap.

Chapter 6

Conclusion

Data and AI are key factors in the ongoing transformation of the business world. The challenge lies in finding suitable projects that align with your Data & AI maturity level. The development of use cases that provide a directly tangible added value for employees and/or customers should be the focus. This paves the way for a positive dynamic that strengthens trust within your organization in this cutting-edge technology and justifies further investment in AI applications.

A structured and well-thought-out approach is greatly beneficial. It helps you to identify and prioritize ideas in the backlog without overlooking your strategic goals and operational challenges. To do this, bring domain experts and AI specialists together in a use case workshop to collect exciting ideas in the sweet spot of business and tech know-how. You should also place value on a structured definition of ideas, to subsequently be able to assess their business value and technical feasibility without difficulty. Thus, an appropriate prioritization and focus on two to three initial AI use cases is quickly implementable and objectively arguable.

By the way, communication with peer groups is considered an enormously important aspect of effectively developing and implementing data analytics and AI initiatives within the company.

So get started, create structures and transparency, and make Data & AI step by step a core competence of your organization.

How Casebase Supports YOU

With Casebase, you can easily build and systematically manage your portfolio for data analytics and AI use cases in a structured way.
A central point here is the identification of use cases, the submission of ideas and the prioritization of your backlog of data analytics and AI use ideas.