Data – the lifeblood of AI today. But when talking about it, people only think of the type and quantity needed to improve the performance of their AI algorithms.

But that’s not enough, especially when it comes to applying AI into business.

Why?

Theoretical AI vs. Business AI

Because unlike in a research or educational environment where the performance of the algorithms is the most important factor, businesses need to seriously think of the ROI of their AI investments. In other words, they must think of the long-term impact of AI on the entire business.

This leads executives to reframe their thinking regarding data when starting any AI initiative.

Instead of “how much and what type of data do we need to make the algorithm successful”, it has become “how can we acquire data that results in the defensibility of my business?”

Long-term data acquisition mindset

Businesses just need enough data upfront to start an AI project. The initial data will attract users (users can be either customers for a new business project or employees for internal process improvement) to use. These users will then generate more data to train the model, leading to better performance, and so on.

This, in turn, creates a loop.

The stronger the loop is, the higher the business defensibility, because the data collected over time are proprietary and un-copiable by competitors.

Here is what Andrew Ng, a world-class AI pioneer, has to say when it comes to launching an AI product.

“When I decide to launch a product, one of the criteria I use is, can we plan a path for data acquisition that results in a defensible business?”

— Andrew Ng, How to Win in the AI Era, 2017.

So, naturally, the question arises:

Since not all data are equally valuable, how can one decide which data, out of the vast amounts collected in any business, can result in long-term defensibility?

An unlikely answer

Although it sounds very technical, this question has to be addressed by a business team, not a technical team.

This is because data and AI are just a means to drive business value. Business executives have to decide what business value should to be delivered. Having that said, many business leaders can feel lost when it comes to talking about data.

A proven framework to deliver business impact

At Nexus, we’ve developed a business-friendly framework called  the “Data Harvesting Loop,” which aims to help business executives identify the AI opportunities that can result in a defensible and scalable business, allowing them to make better decisions on AI investments.

If you are interested in this framework, please take a look at our workshop, “AI-Enabled Business: A proven framework to drive scalability and reusability with AI” to introduce those interested to this framework in detail.