Assessing Machine Learning

If your company is not good at analytics, it’s not ready for AI

This article first appeared in Harvard Business Review.

Management teams often assume they can leapfrog best practices for basic data analytics by going directly to adopting artificial intelligence (AI) and other advanced technologies. But companies that rush into sophisticated AI before reaching a critical mass of automated processes and structured analytics can end up paralyzed. They can become saddled with expensive startup partnerships, impenetrable black-box systems, cumbersome cloud computational clusters, and open-source tool kits without programmers to write code for them. 

Companies must have sufficiently automated and structured data analytics to take advantage of new technologies

By contrast, companies with strong basic analytics – such as sales data and market trends – make breakthroughs in complex and critical areas after layering in artificial intelligence. For example, one telecommunications company we worked with can now predict with 75 times more accuracy whether its customers are about to bolt by using machine learning. But the company could only achieve this because it had already automated the processes that made it possible to contact customers quickly and understood their preferences by using more standard analytical techniques. So how can companies tell if they are really ready for AI and other advanced technologies?

First, managers should ask themselves if they have automated processes in problem areas that cost significant money and slow down operations. Companies need to automate repetitive processes involving substantial amounts of data – especially in areas where intelligence from analytics or speed would be an advantage. Without automating such data feeds first, companies will never discover their new AI systems are reaching the wrong conclusions because they are analyzing outdated data. For example, online retailers can adjust product prices daily because they have automated the collection of competitors’ prices. But those that still manually check what rivals are charging can require as much as a week to gather the same information. As a result, as one retailer discovered, they can end up with price adjustments perpetually running behind the competition even if they introduce AI, because their data is obsolete.


About authors

Nick Harrison is a London-based partner and co-lead of Oliver Wyman’s Retail and Consumer practice globally. Deborah O’Neill is a London-based partner in Oliver Wyman’s Digital and Financial Services practices.

Assessing Machine Learning