How to Pick Winning Analytics Projects — A Proven 4-Step Approach

We all recognize that Analytics and AI can have a substantial impact in driving revenue, reducing costs, or even revamping business models. However, according to a recent HBR article, around 80% of all analytics and AI projects fail. This number shows that not all analytics and AI projects are created equal.

So how can business leaders identify the best use cases for their unique organizations? In this piece, we share a proven four-step framework to determine which projects will give you the most bang for your buck and beat these failure odds, grounded in decades of experience helping organizations design and implement advanced analytics projects.

Our experience working with companies across diverse business sectors reveals a common challenge: Organizations are eager to embark on their analytics and AI journeys but often find themselves uncertain about where to begin or how to effectively approach these initiatives. This hesitance is frequently rooted in a lack of a clear strategy of how to search for the most relevant opportunities.

As a result, many companies either delay their adoption efforts or proceed without a coherent plan, missing out on the transformative potential and often engaging in projects that end up in failure.

Hopefully, the following framework can help tame these initial uncertainties and guide organizations to confidently navigate their analytics and AI journeys.

Step 1: Align with strategic vision and identify "nuggets"

The initial step involves understanding the organization's strategic priorities and key business economics. Engaging in C-level discussions to elucidate the strategic agenda and challenges is crucial. Additionally, examining the organization's profits and loss statement (P&L) can reveal the metrics where the greatest impact is achievable.

Consider a real-world example from the mobility sector, where our C-level meetings identified customer conversion as a top priority for the forthcoming years. By integrating this insight with B2C sales data and current conversion rates, we illuminated the potential value of this priority. Another notable example involves a European telecommunications company. We delved into the P&L figures to determine the most promising "verticals" for the project, uncovering seven verticals, each with clear baselines for optimizing sales, costs, or capital expenditures.

Step 2: Ideate analytics and AI opportunities using a “Pull” approach

After aligning on strategic priorities, the next step is to identify key business areas for interviews to understand current pain points. These interviews are essential for grasping the key processes, decisions, and challenges faced by the business. With a comprehensive understanding of the business context, one is in a position to conduct analytics ideation workshops.

These workshops are pivotal and should foster an innovative environment. Ideally, they are conducted in person and in spaces that encourage ideation and collaboration. The following steps can help generate a broad list of potential analytics solutions for business problems:

• Identify stakeholder pain points.

• Ideate potential analytics use cases to address these pain points (recognizing that not all challenges are suitable for analytics solutions).

• Identify key metrics affected by the use cases and estimate the potential impact.

• Develop a concept pitch.

Next, benchmark the list of identified use cases during the workshop against known industry best practices by reviewing:

• White papers from management and analytics consulting firms

• Industry journals

• Academic papers

 Step 3: Prioritize ideated use cases

With a list of analytics use cases in hand, it is crucial to provide at least a simple description to clarify the scope and the specific business problems addressed. Each use case should then be evaluated against three key parameters:

1. Expected impact to be generated and in which key metric (e.g., low, medium, high impact on sales)

2. Level of complexity of the use cases (e.g., low, medium, high depending for example if it’s a descriptive, predictive, prescriptive, or automation solution)

3. Feasibility to filter out use cases (e.g., lack of data prevents implementation, use case already addressed, high risks of adoption of the teams, partners)

Returning to the European telecom company example, approximately 90 use cases were identified across the organization. Of these, around 25 were prioritized based on the expected impact versus complexity analysis.

Step 4: Characterize and quantify high-priority use cases

At this stage, the use cases may be categorized into key clusters (E.g., top, medium, and low priority). For at least the top-priority use cases, create a one-page document detailing the following:

1. The pain point or challenge to be addressed.

2. A description of the use case and the methodology steps.

3. A literature review with references to articles, papers, and other sources to demonstrate the use case's relevance, ideally including examples of impact on key metrics.

4. Risks associated with the use case and proposed mitigation actions.

5. Quantification of the potential impact on the target metric.

In the European telecom company, the top 25 use cases were quantified, estimating a potential impact of tens of millions of euros. Revisiting the mobility operator example, we estimated that the top 10 use cases could generate benefits amounting to approximately the same order of magnitude.

Embarking on an analytics and AI journey can be daunting, but with a structured approach, organizations can unlock substantial value. Our four-step framework — aligning with strategic vision, ideating opportunities, prioritizing use cases, and characterizing high-priority projects — provides a clear roadmap to navigate the complexities and maximize the benefits of analytics and AI initiatives.

About the Authors:

Pedro Amorim is a Professor and Faculty Member of Industrial Engineering and Management at the University of Porto, and the Co-founder and Partner of LTPlabs. Holding a Certified Analytics Professional and a Ph.D. in Industrial Engineering and Management, Amorim has 15 years of hands-on experience helping hundreds of organizations around the world identify, design, and implement advanced analytics projects. He is the Co-author of the book, The Analytics Sandwich: Bringing People and Artificial Intelligence Together to Unlock Business Value.

Isabel Sousa Pereira is a Senior Manager at LTPlabs, where she has been instrumental in leading numerous analytics scoping projects. She combines strategic insight from her extensive experience in management consulting at Big 4 firms with a profound understanding of how to harness the power of analytics within organizations. Isabel holds a Master’s degree in Industrial Engineering and Management, as well as an MBA, both of which have fortified her expertise and drive in the fields of analytics and business strategy.