The pursuit of data-driven decision-making can make business leaders starry-eyed about data science, believing that artificial intelligence in particular can instantly transform their business. What’s needed is a healthy tension between data scientists and business leaders around what’s possible and workable for using data to drive key decisions.
The ideal scenario is all parties in complete alignment. This can be envisioned as a perfect rectangle, with business leaders’ expectations at the top, fully supported by a foundation of data science capabilities — for example, when data science and AI can achieve management’s goal of reducing customer retention costs by automating identification and outreach to at-risk customers.
Consider Target, which in the mid-2010s had flat in-store sales and a growing digital presence. The retailer decided to go deep into data science and data engineering capabilities in discrete smaller opportunities such as improving on-shelf availability of merchandise, reducing inventory, and improving operating efficiency. The result was a big boost in profitability for the entire organization.
The coveted rectangle, however, rarely occurs. A more workable shape is the rhombus, which approximates the elusive rectangle, but with skewed angles that reflect the push-and-pull of expectations and deliverables.
Unfortunately, what’s far more common is misalignment between expectations at the top of the organization and the foundation of what data science can realistically deliver. The best mental picture of this dynamic is an inverted pyramid. The wide top reflects the C-suite’s oversized expectations for data science impact. The small point at the bottom represents the data science team’s current capabilities, which are often far more modest and develop over time.
Over the last few years, an automaker, for example, dove into data science on leadership’s blind faith that analytics could revolutionize the driver experience. After much trial and error, the results fell far short of adding anything meaningful to what drivers found valuable behind the wheel of a car.
In addition, the inverted pyramid can reflect a lack of appreciation of just how impactful small improvements can be — for example, slight increases in profitability per customer or conversion rates. These modest gains may seem underwhelming for senior leaders who made major investments in analytics. Applied over a large population of customers, however, those small improvements can yield big results. Moreover, these improvements can lead to gains elsewhere, such as eliminating ineffective business initiatives.
The misalignment around data science capabilities can be exacerbated when the sales side of a marketing agency or consulting firm over-promises what its data team can deliver to a corporate client. For example, many consulting companies today have an analytics team that can help the client do something better, such as predict demand for an existing product in a new geographical market. However, the client team is under pressure to sell its data services; as a result, forecasting expertise is instead sold to clients as using the power of AI to transform growth strategies and go-to-market roadmaps.
Within an organization, leaders who learn about the potential for AI to accomplish wide-scale organizational transformation may expect too much of the in-house data science team. At times, leadership may push the data team to stretch. One Fortune 100 company I’ve studied closely has successfully deployed this thinking by creating a data science center of excellence, while also creating a healthy competitive atmosphere that encourages data scientists to push each other to find the best tools, strategies, and techniques for solving problems and implementing solutions.
More often, however, misalignment is a source of frustration — not inspiration. What’s needed are ways to create better alignment and a more productive mindset. Here are three steps to get there:
- Give a dose of reality—to both sides. Consider the example of a data science team with expertise in building models to improve customers’ shopping experiences. Business leaders may assume that a natural next step is to use AI to enhance all customer service needs. What’s clearly needed is better understanding of what AI can and can’t do. AI and machine learning can provide algorithmic output, but that doesn’t necessarily reveal business solutions or how to proceed. AI doesn’t answer “why” or “how” in most cases — humans need to do that based on AI output. It’s not just a matter of reining in leaders’ expectations. Data scientists also need to understand the reality behind the business leaders’ requests to push them out of their comfort zones and explore what they can deliver to move toward bigger and broader goals.
- Build on past successes and achievements. There is value in small data projects to build capabilities and understanding and to help foster a data-driven culture. The best firms keep expectations modest at first. After executing the analytics projects, they conduct a brutally honest post-mortem of the successes and failures, and iteratively build business expectations at the same time as analytics investment. This avoids the trap of thinking that data science is a single endeavor that can or should solve any and all data questions.
- Let data scientists do the talking. Communication around what is reasonable and deliverable given current capabilities must come from the data scientists — not the frontline marketing person in an agency or the business unit leader. Prior to any contract or project, corporate clients should engage with agencies’ and consulting firms’ data science teams to ensure that the sales team’s promises align with what the data science team can actually deliver. To facilitate that exchange, data scientists must improve their ability to “speak business,” relating specific projects and capabilities to solving business problems.
As business leaders and data scientists gain a better understanding of the expectations, objectives, and limitations of the other, a partnership of mutual understanding will develop. With better alignment and a more productive mindset, there will be more opportunities to use data to improve decision-making and achieve better outcomes.