This blog presents findings from managing AI projects, from identifying initial AI opportunities to piloting the most promising ones, and finally, creating full-scale capabilities that can be used in production.
We are looking at this from the AI consultancy perspective and through a lens of a recent client engagement. Our focus was on one of their divisions, specifically the supply chain management processes from procurement to sales, warehousing, and logistics, with an emphasis on the quote-to-delivery process from a business perspective.
Ideate and Design
This blog is organized into steps that are typically followed in the creation of an AI product. Let’s focus on the first part, the AI prestudy, which has two major steps:
- Ideate AI opportunities - This step involves identifying value creation areas with the business and creating and prioritizing AI opportunities at a high level
- Design AI products - This step involves designing prioritized AI opportunities with regard to AI product features, business value, target users, feasibility, ethics and compliance, among others
Gaining executive support for the prestudy
Acquiring executive support for a prestudy is essential. Typically, the process runs more smoothly when senior executives are involved from the beginning and participate actively in the workshops.
However, in some instances, the project may be initiated by middle management, the Center of Excellence, R&D, HR, or another party. While this is acceptable, it’s crucial to develop a clear plan for engaging the executives who will eventually sponsor your AI Proof of Concepts and production deployments.
Some strategies we have found useful:
- Emphasize that ideation workshops can be an opportunity to educate, inspire, and empower participants in identifying AI opportunities aligned with business goals
- Clarify the prestudy phase's purpose, emphasizing its role in developing an AI implementation roadmap
- Facilitate a smooth cultural shift by providing AI training, promoting awareness, and establishing a shared understanding of AI's potential and limitations
- Address misconceptions and fears about AI to foster productive collaboration
- Communicate the benefits of AI adoption and the risks of inaction, using industry success stories to inspire participants
- Utilize innovative and engaging methods to maintain executive interest during workshops
AI Prestudy and Opportunity Identification
AI and Business Strategy
Identifying AI opportunities can be challenging, as many companies need help knowing where to start. The best way to understand the potential of AI is to begin by understanding your business, i.e. what your goals are and the challenges you are facing. This will help you identify the areas where AI can make a difference.
It is also vital to understand what prohibits your company from achieving its vision and how data and AI could help. Typically it is also worth assessing existing data & AI maturity, the ecosystem, and potential disruption in your industry.
Our client faced a significant challenge in meeting customer delivery times for products that were not readily available on stock and had to be procured from suppliers. Addressing this issue was a primary goal for us, and we aimed to leverage AI to improve this metric
The business understanding is typically conducted through interviews and workshops with the top management (group / division / business part).
You can think about:
Strategic Objectives
What are the strategic objectives and priorities of the business/group in the next five years?
Challenges
What challenges should be solved to reach the strategic objectives? How could data and AI help?
Possible disruptions
What disruptions can happen in the industry regarding product, service, or business model innovation?
Value Creation Areas
A good practice is to summarize the business insights into value creation areas, i.e. areas where the company would most benefit from data and AI. These, then, will be linked to business objectives. They should serve as inspiration (not a prescription) for companies looking to use AI in their problem-solving and value creation.
Â
Value creation areas can be high-level business transformation objectives, such as moving from product sales to solution sales (case from manufacturing), or something more specific such as improving clinical decision support with AI (health technology) or personalization and recommendations (media).
Â
In our example, we further categorized the AI opportunities into three value-creation areas linked to the client’s business strategy and objectives. Later each area was assigned the four most promising AI opportunities. The key is that selected AI opportunities can contribute significantly to the success of the value-creation area in question.
Is it necessary for AI use cases to strictly adhere to corporate strategy?
The answer is both yes and no. Generally, it is essential to align the implementation of a AI use cases with the corporate, business, product, or some other strategy to ensure its success. Failing to do so could result in difficulty obtaining broad organizational buy-in, especially if it does not align with executive objectives and is perceived as optional at best.
Â
Nonetheless, it is important to allow room for innovation. Using AI to approach things differently can result in improved outcomes and present new business opportunities that may have gone unnoticed otherwise.
Â
So, the diplomatic answer is that while creating AI use cases, it is vital to anchor them in the corporate strategy framework, while also leaving room for AI’s capability to exceed expectations.
Identifying AI Opportunities
Ideation workshops are an essential part of the AI product design process. They are used to create ideas for new AI products and to prioritize among them. Each opportunity is evaluated in terms of feasibility and business impact.
Â
Whereas the value creation area identification is a more top-down approach, AI opportunity ideation often starts bottom-up. Usually, the sweet spot is somewhere in between, meaning the project-level AI use opportunities are combined together and contribute to the bigger vision (instead of working against organization’s overall goals).
For example, an organization wants to become more customer-centric (a high-level goal). Project-level AI opportunities in this context may encompass customer segmentation and the prediction of conversion and churn rates. Individually, these initiatives may seem insignificant, but in conjunction, they advance the organization’s overarching goal of becoming more customer-centric.
In general, there are several methodologies for identifying AI opportunities. As a starting point, we typically use the following quadrant that applies to different industries relatively well, although the perspectives need to be revisited sometimes. For clarity, we also collected examples of potential methodologies useful in these areas.
Operational excellence aims to optimize the efficiency and effectiveness of internal operations, addressing questions like
- How can we replace manual or repetitive processes?
- How to reduce costs and save time?
- How to get better business insights?
Â
Digital customer experience is primarily focused on enhancing the customer journey. To achieve this, businesses may pose the following questions
- How to engage the customer more effectively?
- How to eliminate the barriers to buying?
- How to better understand customer needs?
Â
The enhancement of existing products and services using data and AI is a key aspect of product and service innovation. In this context, companies may consider the following questions
- How to leverage data and AI in product and service development?
- How to incorporate data and AI into existing products and services?
- What new capabilities or features could be enabled with data and AI?
Â
Growth and business model innovation seeks an understanding of how to grow and renew the business. It answers questions such as
- How to increase revenue?
- How to expand the customer base?
- How to create fresh offerings, services, and business models that incorporate data and AI advancements?
AI use cases around operational excellence
We chose the process analysis approach because this study focused on operational excellence and was tightly related to their internal business processes. Typically, the business value in this area comes from increased efficiency, cost reductions, and reduced manual work.
Â
A simplified version of such a process map is presented below:
After a handful of interviews and workshops with the subject matter experts and business executives, we had tens of AI opportunities mapped in their business processes to solve the identified pain points.
Prioritizing AI opportunities
Once you have identified many AI opportunities, you need to put those in order. The simplest way is to prioritize them on a high level based on their business value and feasibility. Most consultancies use the following matrix.
Business impact and value
Assessing the business impact and value is typically straightforward, particularly when viewed through the lens of revenue generation or cost savings. However, the impact can also extend beyond financial metrics, and instead be aligned with the company’s strategic objectives or key performance indicators, such as customer satisfaction.
Â
Naturally, determining the impact requires input from relevant business stakeholders.
Feasibility
The feasibility, on the other hand, is more difficult to quantify. Factors that contribute to feasibility include for example the availability of data, complexity of machine learning solutions, readiness of business processes, and even the organizational culture.Â
Â
Determining feasibility often necessitates insights from various types of stakeholders, including domain experts, data scientists, and business leaders.
Further prioritization - RICE scoring model
After completing the initial high-level prioritization, it is beneficial to conduct additional workshops that examine opportunities from various perspectives. One effective framework for this is the RICE scoring model, commonly used in product management.
Â
The subsequent round of prioritization workshops can be particularly valuable when the conversation delves into discrepancies and uncertainties surrounding the identified use cases.
Â
This can help refine the use case design, potentially unearthing novel perspectives that had not been previously considered. Through this iterative process, the organization can better identify and prioritize the most promising AI opportunities.
AI use case design
AI use case design phase attempts to specify the most promising opportunities in more detail.
Â
In our example, selected use cases were analyzed and designed in more detail. While there are canvases available for free that facilitate the analysis and definition of AI/ML use cases, they can be overly technical and lack a business-oriented perspective.
Â
To ensure a comprehensive approach, we like to examine the opportunities from multiple angles. This includes gaining a deep understanding of the business problem, identifying potential impacts on business processes, specifying the necessary data requirements, and establishing clear objectives for the AI solution, including how to measure its success. Through this multifaceted approach, we can ensure that the use case design is well-rounded and aligned with the organization’s goals.
Even high-level AI use case analysis requires a diverse set of skills. You need to have
- Business professionals and subject matter experts who can provide a detailed understanding of the business problem, including its impact on existing processes and how success will be measured
- Data experts who are knowledgeable about the available data, its quality, storage, and appropriate format
- IT experts who understand where data is located and how to access it
- AI experts who can define the AI task at a sufficient level, including input parameters, predicted output, model selection, prediction horizon, and model inference, among others
AI use case portfolio and roadmap
Having a comprehensive list of qualified AI use cases alone is insufficient. To drive execution, it is crucial to structure the use cases into an AI use case portfolio and develop a roadmap.
Â
In our example,after completing the intensive five-week opportunity mapping phase, we summarized the prestudy results and presented them to the executives. Along with the presentation, we shared a detailed roadmap and budget that highlighted low-hanging fruit opportunities and those that would bring substantial long-term business value.
Â
By structuring the AI use cases into a portfolio and roadmap, organizations can in later stages further prioritize opportunities, allocate resources, and systematically execute their AI initiatives. This approach ensures that organizations are well-positioned to achieve their goals and reap the benefits of AI.
AI PoC candidate selection
The initial findings and PoC candidates were discussed with the project team of 15-20 persons, from SVPs to middle management and individual contributors.Â
Â
The core is to keep things as simple as possible and in business language. Highlight the what (is in for me) over how (it is being done). Never too often, AI scientists tend to focus on the technicals while executives thirsts for business benefits.
Â
The findings were also presented on the client’s strategy day for the extended management team to get more buy-in and support for the implementation phase. We had only three high-level presentation slides to showcase in our 15-minute timeslot.
Â
Eventually, five AI cases were selected for piloting. Some were different proposed by us because of data availability, client resourcing, and other factors.
What's next?
In the subsequent segment of this blog series, we will shift our focus to the AI piloting stage. Here, we will delve into two distinct use cases – one that yielded exceptional results and another that fell short of expectations. Through these examples, we will gain a deeper understanding of the underlying factors that contributed to their outcomes.
Â
Stay tuned!
Key takeaways from the prestudy phase:
- Ideation workshops should not only generate AI opportunities but also educate, inspire, and empower participants to generate innovative ideas
- Executive support is crucial at an early stage. Ideally, a senior executive with ample resources should drive the initiative forward
- Ensure that business people, subject matter experts, and data experts are available and booked well in advance
- Unicorns with expertise in business processes, data, and technology can add significant value to the ideation workshops
- Introducing AI involves a cultural shift that must be managed. Resistance to change is inevitable
- Tie AI use cases to corporate strategy but leave room for innovation outside the strategy to explore new possibilities. Flexibility is key to capturing opportunities to do things differently
