Over the last few years, business leaders have invested millions into setting up AI/data science teams to gain a competitive advantage. Some AI initiatives have resulted in measurable benefits, but many haven’t. Initial exuberance among business leaders has given way to skepticism.
In our experience collaborating with Fortune 500 enterprises at TheMathCompany, we’ve observed that despite having sophisticated AI tools and star-studded data science teams on their side, weak links in strategic and operational aspects have deprived organizations of deriving meaningful value from AI.
Based on our work across industry verticals helping organizations go through analytical transformations, I recommend five best practices to ensure tangible and sustainable value from AI.
1. Ensure AI priorities are closely aligned with business priorities.
What is your organization’s one-year and three-year business plan? What is your five-to-seven-year vision? Does the AI adoption road map reflect these priorities? If you want to get real value from AI, the AI/analytics function should be led by a senior person who has a say in organizational priorities and who ensures AI investments are aligned with them.
Taking up an AI-driven marketing overhaul when your company is going through a major supply chain transformation is like taking a pill for the heart when you have a kidney problem. While it’s good to have a healthy heart, it should not be prioritized over the kidney pill. The most important aspect of making an AI project successful is identifying the right problem. And the most important attribute of the right problem is that it is in line with business goals.
2. Invest in data.
Here is the hard fact: Algorithms that took your data scientists months of effort to develop are fast becoming a commodity. Today, somebody with a foundational knowledge of data science and no programming skills can build a very good model with just a few clicks. This means your competitor can also access the sophisticated algorithms that you employ, again with just a few clicks.
So, what is the differentiator? Data! The one who has the best (most granular, clean, timely) data has an edge over the rest. For that reason, it is imperative that your organization find a data strategist who can identify the most impactful data elements and ensure they are procured, so you can develop better insights than the competition.
3. Prioritize quick wins.
AI initiatives have to be closely integrated with business initiatives in order to create maximum value. For this, AI has to move at the speed of business. Invest only in bite-sized minimum viable products (MVPs) that the business can start using within a quarter or so (even complex applications can be broken into series of MVPs). This way, it stays relevant, showcases value and exposes itself for real-world feedback.
Three dimensions suggest whether an AI use case is a quick win:
- Value that it is estimated to generate at full adoption (note that value doesn’t necessarily need to be a dollar amount, but it has to be measurable).
- Technical complexity in achieving the output and, hence, the time and effort involved.
- Sponsorship from a senior business stakeholder who provides not only the budget, but also time and guidance to the AI team, and who also iterates with them.
4. Invest in translators.
Let’s bust one more myth. What is the first genre of talent you should be recruiting when you set out on an AI path? Data scientist? Think again.
Much of data science today can be accomplished by clickable automated machine learning (AutoML) platforms. Or you can hire consultants to fill the gap. What is more important for your organization is a group of analytics translators — cross-functional professionals who act as liaisons between business teams and analytics teams.
The job of a translator is twofold. On one side, they work with business teams, educate them on what is and isn’t possible, and help them identify the right use cases. On the other side, they work with the data science team to translate the business problem to an analytical problem, run the projects with an MVP approach and interpret the business meaning of analytical results.
Once the solution MVP is ready, they work again with the business to validate and internalize the solutions, with appropriate change management interventions. As such, translators need to have hands-on knowledge of your business, deep relationships across the organization, a working knowledge of analytics and cultural empathy for the analytics team.
Translators hold the key to unlocking value from your AI investments. Invest in them, and choose them carefully.
5. Take a course.
A lot of companies don’t realize the value of AI investments because there is a lack of understanding and trust. As a leader, it is important that you (and your direct reports) understand the “why” and “what” of AI, so you can create a top-down AI culture. You don’t have to be a data scientist, but you should be aware of what data science can and cannot do. Awareness leads to better choices in people, processes and technology investments and can lead to better outcomes.
So, if you haven’t already, now is the time. Log on to Coursera (or any other MOOC) and take an introductory AI course, and mandate the same for your middle management.
If your analytics initiatives have not shown results, it is likely that you have fallen short on one of the above practices. Execute a root cause analysis, and take corrective measures so future initiatives have higher chances of success.
What you have seen here is a cross-sectional view of five aspects that matter when it comes to deriving value from your AI investments. However, if you are wondering where you should start your journey, my recommendation is that you identify a team of two to three translators who report to a senior person with an analytics mandate. The goal for this team is to implement a few pilot projects and showcase success stories to the rest of the organization. This will create momentum for the transformation.