Consider the following hypothetical scenario at your business: In anticipation of a new product launch, you have commissioned focus groups to test the market reaction to your product and messaging. When the research has been completed and the results delivered, you gather leadership from across the company and eagerly dig in, only to find that, surprisingly, the entire report is written in a language that no one in the room understands.
The simplest way to address this situation would be to find someone at your company who can translate the report into a common language. When the translation has been completed, the translator can hand off the accessible information and resume his or her regular responsibilities. The team of leaders can now use that report to make the decisions they need to make. It wouldn’t make sense to ask the translator to make company-wide decisions about the new product simply because they processed the information.
This hypothetical scenario is becoming a reality for many data teams working on artificial intelligence (AI) projects today: They collect loads of information and build advanced prediction models with it, but instead of making that data available to teammates, they are charged with making the business decisions themselves. Because they speak the language of AI, they’re in charge of the entire process. Since the work involves heavy use of disparate data sources, distinct analysis tools and languages that aren’t easily learned, these complex projects run in total isolation from the normal decision-making workflow. They are handled end-to-end in the data science realm, excluding other teams.
For the leaders of business teams, data from exciting AI and machine learning technology is important, but it isn’t something they can model on their own or analyze easily. Solving this problem by translating this valuable base of information into something those leaders can use is one of the most important issues businesses should be addressing today.
The answer to this problem starts with the data team. These are the employees who speak the languages of complex data most fluently, so they should be the ones who read it first. In fact, many data teams today are making progress in bridging the gap between complex data and their own analytics tools. But there’s still a significant barrier between what they can analyze and how it reaches the final decision maker. As a result, data teams are being asked to do it all. Once they’ve read the data and created models based on the information, they’re responsible for making predictions and recommendations back to the business teams. They’re doing both jobs: translating and making the final business decision.
This is a great way to make sure that the person who understands data the best is making decisions from that data, but it also runs the risk of removing the actual business experts from the process. In the same way that you wouldn’t have the translator from our hypothetical example making the final decisions with the translated text, a data team shouldn’t always be making decisions for other business teams. Speaking the right input language doesn’t necessarily mean that someone is the best-suited individual to produce business output.
The solution to the question of how to use AI to turn data into business decisions is to let your data team create models that the business leaders can understand, then give those leaders access to the translated data and let them use their expertise to dig in further to make decisions. This process lets both teams realize the value of their specific skills.
Empowering citizen data scientists is a win-win for proactive data teams that want AI to infiltrate other parts of their business. Giving these teams access to prepared datasets will give them the freedom to explore their instincts within a safe analytics environment. When data from predictive models or machine learning is presented in a format that’s familiar, it will truly remove the final barrier to making that data fully usable throughout the business.
This process ultimately eases the burden on the data team by asking them only to focus within their specific area of expertise. It’s a great way to make an entire organization more fluent in reading data without asking a wide swath of the company to learn the difficult new skills associated with predictive AI models.
Having a data team that can speak the language of AI is valuable, but having a team that makes that language accessible to other members of the organization is even more important. Your data professionals can be best utilized as a common resource rather than a siloed function. Instead of having the same team do both the translating and the reading of the data, separate the processes and have the data team do the modeling — the translating — and then pass it off to other teams to do the analysis — the reading.
The future of realizing value from AI isn’t unclear or far away. There are people at your company right now who know how to speak that language; they just need to be used the right way. You can start by creating clear channels of communication between the data team and other employees who will benefit the most from their data models. Once everyone is speaking the same language, you can start making smarter AI-based decisions today.