top of page
  • Writer's pictureEitan Netzer

The Misconception about Data-driven Insights

"Insight: (the ability to have) a clear, deep, and sometimes sudden understanding of a complicated problem or situation"

Organizations often say: "AI will provide us with insights". It seems that everybody these days expects AI and data analytics systems to draw insights from big data. However, AI does not have its own knowledge and cannot generate insights from data. AI by itself does not know how to "connect the dots" to draw valuable conclusions.

You have two ways to generate insights from data using AI:

  • Direct insights: Specify which insight you need. For example, ask AI to predict how many products will be sold to know how many units to produce. We call it "supervised learning", where you instruct the computer to find the exact information you expect.

  • Indirect insights: Instruct the computer to identify trends and patterns in the data. The computer will do the heavy calculation work. It will provide the findings from which an analyst or a domain expert will draw conclusions and provide actionable insights.

Ask smart questions

Big data contains enormous amounts of useful information. But unless you know what you are looking for, search in a smaller pile, and ask the right question, you are not likely to find any valuable answers.

Do you expect the computer to ask the questions for you? Sorry to disappoint you. For a computer to identify the right questions, you need the budget and computing power of Google…

How to know which questions to ask?

The questions to ask AI need to address a business problem. If you do not know which questions to ask, here are three tips for finding the right questions:

  • Consult with an AI expert who can share with you some of the common questions businesses ask (as questions tend to repeat themselves across departments and industries). For example, marketing executives often want to identify the most profitable customer journey or predict and prevent customer churn. These, however, are still very broad, and you may want to get even more specific and ask specific questions that lead you to possible solutions.

  • Check what questions the company's Business Intelligence (BI) team asks. For example, the BI people of a retail chain may have a dashboard that reports the daily sales of each store. Instead of asking the BI system to report the daily sales, ask the AI system to predict the daily sales.

  • Use an AI tool to scan the data and identify trends and patterns. Provide the output to a Data Analyst who is highly familiar with your business domain. Based on the trends and patterns found in the data, the analyst will ask the right questions and gain substantial insights. (Notice: a data scientist is NOT a domain expert. It might take a data scientist 9-12 months to learn the domain and provide valuable insights).

Making AI accessible to BI

At CoreAI, our approach is to make AI accessible to the BI team without depending on a data scientist. We build a pipeline of questions, and our AI solution queries the data. The domain-expert analyst can then take the output, analyze it, and extract valuable insights that the organization can turn into actions and business results.

64 views0 comments


bottom of page