How to Create a Compelling AI Use Case Roadmap

Hi there,

Have you ever spent months on an AI project that never got off the ground? I've been there. Four months in, my "strategic use case" was dead in the water – no implementation, no plan B, just wasted time and resources.

The reason? There was no use case roadmap.

A well-crafted use case roadmap is a critical tool for any organization to gain value from AI. In today's newsletter, you'll discover different use case types and learn how to combine them for clarity and direction instead of chasing the hype.

What is an AI Use Case Roadmap and Why Bother?

An AI use case roadmap is part of an organization's AI strategy that defines the specific AI use cases that will be implemented, together with their goals and associated timelines.

By "AI Use Case" I mean a specific scenario where artificial intelligence solutions can be applied to solve a problem or improve a process within an organization.

In some organizations, this use case roadmap is the AI strategy.

Creating such a use case roadmap can be challenging.

Here are some common reasons why creating a good use case roadmap is so hard:

• Complexity: Juggling multiple use cases, data types, and technologies can be overwhelming.

• Uncertainty: The fast-paced nature of AI makes project planning a challenge since the landscape is always shifting.

• Resource Allocation: It's tough to predict which use cases will give the best ROI.

• Time Constraints: The pressure to deliver quick results may make planning seem like a luxury.

• Lack of Clarity: Defining the potential impact or feasibility of a use case can be difficult.

Don't panic! Remember, your roadmap is a high-level guide, not a blueprint. And it should be designed to evolve.

Embrace its flexibility, and it'll become a less daunting, more helpful tool.

As the old saying goes: "Planning replaces coincidence by error."

I think that's a good line to live by.

Step 1: Assess Feasibility and Business Impact of Use Cases

When crafting your AI roadmap, the first step is to assess your use cases based on two core factors: feasibility and business impact.

Feasibility refers to how easy or hard it would be to implement a use case, considering your data, infrastructure, expertise, budget, and other constraints.

Business impact, on the other hand, gauges the potential value the use case could bring to your organization.

(Check out chapters 1 + 2 of my book for frameworks to assess feasibility and business impact)

Step 2: Identify Use Case Categories

After assessing your use cases, you need to map and rank them based on their scores.

This process helps you categorize your use cases into four broad types:

Champions: High Impact, High Feasibility

These are the dream use cases for any AI project.

Champions have both high impact and high feasibility. They're your star players, set to bring significant value to your organization while being reasonably straightforward to implement.

Prioritize these use cases, as they are instrumental in driving your AI roadmap.

Examples: Customer churn prediction, fraud detection, next best offer

Quick Wins: Lower Impact, High Feasibility

Quick wins may not bring a massive business impact, but they're just as vital.

With high feasibility but lower impact, these use cases are typically less complex and quicker to implement.

They serve as excellent showcases of what AI can do, and they're great stepping stones toward more complex projects.

Examples: Customer segmentation, sentiment analysis, Q&A Chatbots

Research Cases: High Impact, Lower Feasibility

These use cases promise high impact but are not readily feasible.

Whether it's a lack of data or some technical constraints, these use cases need more time and resources to materialize.

However, given their potential value, you should keep them on your radar.

Examples: Predictive maintenance, personalized medicine, autonomous vehicles

Reassess Later: Lower Impact, Lower Feasibility

Don't discard these just yet!

Use cases falling into this category may seem less appealing due to their low impact and feasibility. But as technology evolves, what seems unfeasible today might turn into a quick win tomorrow.

For example, voice recognition technology was once considered a reassess later use case, but thanks to advancements in audio processing and LLMs it's now almost a quick win.

Revisit these cases periodically to check if they've become more feasible or impactful.

Take a moment to reflect. What AI use cases are you currently working on and how would you categorize them? What would you do next?

Step 3: Build a Balanced Roadmap by Mixing Use Case Types

Creating a compelling AI roadmap is all about understanding how to mix and match different use case types to strike a balance between ambition and pragmatism, between quick wins and long-term goals.

Here are a few tips to achieve that:

1. Mix Champions and Quick Wins: While it's essential to focus on high-impact, feasible use cases (Champions), don't overlook the Quick Wins. These less complex projects often have shorter development times and a high probability of successful implementation, which can keep your team motivated and demonstrate the value of AI to stakeholders - especially when you’re just starting out.

2. Keep an Eye on Research Cases: These are your long-term goals. They might not be feasible right now, but their potential impact is high. Regularly revisit these cases to check if technological advancements have made them more feasible - ideal for more mature AI organizations.

3. Don't Dismiss "Reassess Later" Use Cases: These are low on both feasibility and impact, but that doesn't mean they aren't valuable. They often provide great learning opportunities or yield unexpected benefits. Also, as technology changes, these use cases can turn into quick wins. Keep an eye on them throughout your journey!

4. Identify Common Data Sources: When choosing between use cases, favor those that utilize the same data sources. It helps to streamline your efforts and reduce complexities.

5. Maintain a Flexible Approach: Remember, your roadmap is a guide, not a strict plan. Keep it adaptable to accommodate new insights, changing business goals, and technological advancements.

6. Build a Compelling Vision: Include even those high-impact use cases that might still be far from technically feasible. These 'moonshot' goals can inspire your team and align other, less complex use cases towards a common objective.

Creating an effective use case roadmap is both a science and an art. It's about understanding your use cases, prioritizing them effectively, and maintaining a balance.

A well-crafted roadmap can guide your AI journey, ensuring your projects are aligned, organized, and set for success.

Conclusion

Crafting your AI use case roadmap isn't a one-time task, but an ongoing process.

Start by assessing your use cases for business impact and feasibility. Categorize them, then balance your roadmap with a mix of Champions, Quick Wins, and Research Cases. Remember, 'Reassess Later' cases might soon turn into tomorrow's Quick Wins.

Your roadmap is flexible. As new insights and technologies emerge, revisit and adapt it. The key to AI success lies in iteration.

Don't just create a roadmap - evolve it.

That's it for today! I hope you enjoyed this article. If you did, feel free to share it with a colleague, spread the word, or just drop me a line!

I'd love to hear your thoughts.

Until next Friday!

Tobias

Want to learn more? Here are 3 ways I could help:

1. Read my book: Improve your AI/ML skills and apply them to real-world use cases with AI-Powered Business Intelligence (O'Reilly).

2. Book a meeting: Let's get to know each other in a coffee chat.

3. Follow me: I regularly share free content on LinkedIn and Twitter.