The organization of the future will be data animated

(I was talking to Alan Murray, Editor at Fortune, about data animated organizations and decided it was time to hit the publish button on this post.)

“Artificial intelligence” and “big data” are over-hyped terms, but their promise is real. The past four decades saw the rise of at least three major technological eras: the PC, the Internet, and mobile.  Each started out as toys for nerds, but grew to impact the world. We are now on the cusp of a new era, fueled by data and run by machine intelligence.

The term “data driven” doesn’t do it justice.  Most organizations are already data-driven in the sense that their decision-making is supported by data. But data still doesn’t run the organization.

For the companies embracing this new era, I propose a new term: data animated.  Like a self-driving car, data animated organizations are better equipped to steer clear of disaster and optimize their routes.  Their use of machine intelligence doesn’t exactly mean that models will run the world, as Steve Cohen suggested earlier this year, though, to be sure, models will be a critical part of data animated organizations. Rather, the successful data animated organization will use machine intelligence to maximize the potential of the vast amounts of information now being collected every second around the world. It will do so by employing three essential tools: a data refinery, digital twins, and a new interface.

Data animated organizations require data to run, which means a data refinery must be at their core.  I’ve written elsewhere about data refineries and how they work.  Essentially, the refinery gathers data from inside and outside of the organization, cleans it up, and puts it all in one place.  A data refinery isn’t just a data warehouse, it is a system that readies the data for model building and then gives organizations the computational power and AI tools to accelerate model creation.

Models sit on top of a data refinery and further distill information so that useful insights that pertain specifically to an organization can be collected. From these insights, the models ultimately build digital twins of an organization’s processes, operations, supply chains, and markets. How might that be useful? Imagine a mining company that builds a digital twin of its supply chain.  Calculated predictions and trend observations conducted by the digital twin (based on real-time, global data that’s constantly streaming into the refinery) can help the company answer questions about whether or not to open a mine on a particular site, or where to ship ore based on demand factors.  

Digital twins can be thought of as an A/B test for companies that operate in the physical world. We hear about this sort of testing a lot with digital companies like Netflix, which constantly experiments with new show recommendations to refresh user preferences, or Facebook, which is running thousands of experiments at any given time so it can make product decisions that better optimize usage and revenue. For companies operating in a physical instead of digital domain, a digital twin provides the power to A/B test scenarios in the real world, and arms decision-makers with better information.  

The ultimate output of a digital twin is a prediction of the future, within a confidence interval.  Sometimes these predictions are fed into other AI platforms to make automatic decisions, but in many cases, humans will still have an important and evolving role to play. Human decision-making will be empowered by stronger data, which will be harnessed through a new interface.  

The nature and amount of data produced by an organization’s models will be very different so the data animated employee will require a new kind of interface to make use of it all. Put simply, we humans can’t handle or use all of this information as it stands. Imagine a grain buyer, accustomed to receiving weekly production estimates at the state and national level, who is suddenly inundated with daily forecasts for every county in the country.  That amount of data - even after a data refinery does its work - requires a specialized interface to visualize the information, find patterns, and translate them into something the buyer can understand and use.

“Interface” implies a presentation of data on a screen, but it doesn’t have to be. Interactions with an interface might include commands or questions like, “Show me all of the countries where production is down by more than 10% from the previous 10-year average,” or, “Is there any correlation over the past 30 years of soy production and deforestation in Brazil?”

The learning machine won’t just look up data - like Siri or Alexa do now - it will generate a reasonable answer in context. Responses will include insights, suggestions, and observations of anomalies that could impact the organization.  It will handle a deluge of data and become a domain expert.  It will ask other machines when the full answer lies outside its scope.  It’s not a spreadsheet, or an SQL database, or even Tableau. Engaging with the interface will be like having a conversation with Data from Star Trek: The Next Generation.

These three pieces of technology - the data refinery, the digital twin, and a next-generation interface - are the most critical tools for the data animated organization, and their development and use is rapidly accelerating.  Strategic companies are already making investments in this future reality and experimenting with new types of data, models, and modes of interface interaction.

Ultimately, the era of the data animated organization will see these self-driving companies free up their people to rapidly experiment with bold new strategies thanks to a clearer view of the world around them. They will test, define, and change business models based on insights and autonomy from operations that allow them to be proactive about the future. As they do, we’ll see stubborn stalwarts who cling to a reactive, data-driven strategy rapidly fall behind and find themselves scrambling to catch up at the end of the adoption curve, much like the Yahoos and Netscapes of the last generation.