As the confetti from New Year’s celebrations is swept away, and resolutions fill the air, Analytics Leaders should place one goal at the top of their lists – preparing for the AI revolution of 2024. With the advent of Artificial Intelligence (AI) and its increasing influence on business analytics, it’s time to gear up, get ready or get left behind.
Looking Forward: 2024
AI is going to power the next generation of Business Intelligence tools. We’ve seen a sea change in the industry in just a few short months. Tableau Pulse is now in beta, giving a new UI to metrics in Tableau. ThoughtSpot Sage has gone live, empowering natural language search and answers over your data. Microsoft Copilot is empowering new capabilities in Power BI and Fabric. That’s not to mention all the other competitors and start-ups in the space. It’s an exciting time with accelerating AI adoption.
However, the road to AI adoption is not without its challenges. Many “last-mile” problems will provide significant hurdles to AI’s impact. A “last mile” problem, a term borrowed from the telecommunications field, refers to the final leg or step in a journey. That last mile can often be the most challenging. For example, airplanes and trains provide an easy and efficient way of transportation from one city to another. We need one solution that scales for everybody moving from one town to another. It’s direct and relatively simple. However, if our task includes getting everybody to their homes, we need a network of roads and likely multiple forms of transportation (public transit, cars, bikes and walking). In short, the last mile is the many small solvable problems that are complicated because of their sheer number. You address them over time with a robust strategic plan.
The latest models, like Open AI’s GPT-4 are the planes of AI — powerful, general purpose and transformative. To get us home, however, AI will need access to relevant data about our business, context to make sense of that data and a way to deliver it in a way relevant to the company. These are the last-mile problems that we can start tackling now.
The Last Miles
The first problem to tackle is having clean and accessible data. Good data is a prerequisite for leveraging AI. It ensures that AI models have accurate and high-quality information to learn from, thus increasing their precision and reliability in predictions. Depending on your purview, it can mean cleaning up your BI data sources and revisiting your BI Data Governance strategies. However, suppose you’re looking at a complete company strategy. In that case, it means making sure you’ve developed the proper data architecture strategy and have begun building out the business logic and metric layers in a way that can be shared easily between tools.
If you haven’t already, you’ll hear a lot about semantic and metric layers this year as they help us address the second last-mile problem, giving AI the context of our data. In short, they’re a way to get the info often stored in our analysts’ minds or our business in a centralized place, allowing it to be shared, managed, and updated. The importance of the semantic layer becomes even more evident when we delve into work done by researchers from data.world, which shows that knowledge graphs help improve the accuracy and reliability of answers from LLMs.
This makes intuitive sense — AI will be better at using our data if it understands the context of our business. Just like training a new person in a highly technical company, there are a lot of rules, acronyms and definitions that we need to transmit. In some ways, you can think of data catalogs and semantic layers as your onboarding program for your new AI coworkers. Like any good onboarding program, doing it before your coworkers start their jobs is more effective.
Similarly, identifying the best ways to measure your business becomes paramount. These facts will feed into AI and produce recommendations and guidance. Removing the noise in our metrics to focus on what matters most is another tool we can use to make AI more impactful. As such, avoid vanity metrics and focus on what drives value in your business. So, in 2024, it’s essential to revisit your Key Performance Indicators (KPIs) to ensure they correctly reflect operational and strategic objectives.
The final last-mile problem for AI is delivering the results in a way that impacts the business. Defining the delivery patterns now can make a difference in how you put AI into production. The right solution will depend on your business context (and often, you’ll need to leverage multiple). Still, the options include email distribution, chat application posts (such as Slack or Teams), or a centralized analytics portal. Tools such as Tableau Pulse focus on making it easier to deliver analytics where and when you need them, with a considerable focus on email and chat application distribution. The important thing is for you to start discovering the patterns that work for your company, so begin testing and iterating on your processes now.
New Year, New You
So, as we enter 2024, let’s make a resolution to prepare for our new AI coworkers. Let’s solve our last-mile problems and unleash the full power of AI with clean data, documented business logic, impactful KPIs and effective delivery. By doing this, we stand to reap the significant benefits AI promises in the coming years. As you begin this journey, we’d love to help. What’s your priority for 2024?
