Move your organization past the "I'm smarter than a data system" myth to a data-driven culture by focusing on logical reasoning, credibility, and emotion.
An overemphasis on logic often thwarts the efforts of enthusiastic but naive leaders who are trying to build a data-driven culture; however, an underemphasis on logic can be just as damaging to success.
Just because you and your data scientists immediately see the logic and reasoning for using data to drive decisions, don't assume the rest of the organization does. In fact, it's likely that most of your organization will have a hard time blindly trusting your data. Credibility and emotion play a large part in moving them to your side, but logic has a role as well.
Aristotle's rhetorical triangle teaches us that persuasion comes in three forms: ethos (credibility), pathos (emotion), and logos (logic). Although I encourage you to focus heavily on ethos and pathos, without logos, your organization may have a tough time making the transformation. Data-driven concepts must make logical sense before your organization fully adopts them.
Humans vs. machines
To build a logical foundation for your data-driven culture, you must answer the question, "why is it better to trust a data system over my own judgment?" It is very common management advice to start with some sort of analysis, and then for a final decision, use your best judgment. Is this good advice? For a data-driven culture, I would say no.
In a data-driven culture, the data actually drives decisions -- there's no human intervention required. This will be difficult for many in your organization, because it's generally accepted that humans are smarter than machines. But what exactly does that mean? Some education will help your organization understand the logic of trusting your data systems.
Let's start with memory. Is it likely that a human can recall stored information as well as a computer? With only a few seconds to contemplate this question, most people will arrive at the obvious answer. You can store an obscure piece of information in a database and keep it there for decades, with very little chance that it will forget anything. Can you do the same? No way.
Another thing that's hard to contest is that a computer can calculate numbers and crunch through formulae faster than a human. Even an ancient calculator can do long division much, much faster than any human, right?
So, in the areas of retrieving stored information and quickly performing mathematical calculations, machines clearly have humans beat, but what about inductive and deductive reasoning? This is of course what separates us from all the other living creatures on the planet, let alone our silicon-based aspirants. Even the most advanced artificial intelligence pales in comparison to any human's basic reasoning skills. That's why it's best to stay away from convincing people in your organization to blindly trust advanced systems that can't be logically explained.
An expert system is an easier sell because it's based on storing vast amounts of expert data and using clearly understandable rules for retrieving the data. Asking your organization to trust a neural network is an entirely different thing. If you're faced with this challenge, it's best to double-down on the other two legs of the rhetorical triangle.
SEE: Big Data Primer for IT Pros (Tech Pro Research)
How much logic is required
The degree to which your data-driven concepts must make logical sense to your organization depends on how well you do with the other two legs of the rhetorical triangle: credibility and emotion.
Credibility is a general belief that the data and subsequent analyses are believable. Base data is credible when you have high-data quality and you've done your job to educate the organization on certain nuances in the data (e.g., expected variations). Analyses are credible when they're proven (in the eyes of the skeptic) to be right.
You'll run into challenges with analytic credibility when you deploy advanced, black-box systems into your organization, like neural networks or genetic algorithms. Since you cannot easily explain how they work, people in your organization will have to experience them working to build credibility. When they do start working, the need for logical reasoning goes down -- people will just trust that they work.
The other way to lessen the need for logic is to use emotion. Between logic and emotion, the latter is the more powerful persuader.
If you have terrible events in your organization's history that you can blame on the lack of analytic prowess, this will compel your organization to consider a more data-driven approach. I've worked with companies that have harmed and even killed people because they weren't paying enough attention to their data practices. I've also worked with companies facing the loss of huge revenue streams for the same reason.
These are strong motivators for trying something different, even when the organization can't find a logical reason for relying more on data to make decisions. When you combine this with the credibility aspect previously discussed, it really reduces (but doesn't eliminate) the requirement for logical reasoning.
To move your organization to a data-driven culture, there must be at least some logical reasoning that your workforce can anchor on. The good news is that it takes very little time for most people to understand that a data system can both retrieve stored data more effectively and perform mathematics more expediently. This will help move the organization past the "I'm smarter than a data system" myth. However, never ignore the importance of credibility and emotion to buttress your logic when it falls short in your persuasion efforts (and it will).
As long as you have the other two bases covered, a little logic goes a long way, so don't overdo it.