Prediction Machines by Ajay Agrawal, Joshua Gans, and Avi Goldfarb - A Summary

You’re receiving this because you signed up on our website. Want to unsubscribe? Just reply to this email with the words “no thanks.”

First-time reader? Join other business leaders for free.

Listen to this post here.

Sponsored by

Want to sell more?

Instead of paying salaries, many companies hire commission-only reps—experienced sellers who get paid only when they produce.

Post your opportunity on CommissionOnly.com.

Thousands of motivated reps are actively looking for performance-based roles.

Author: Ajay Agrawal, Joshua Gans, and Avi Goldfarb

Ajay Agrawal is a professor at the University of Toronto’s Rotman School of Management, Joshua Gans is a professor of strategic management, and Avi Goldfarb is a professor of marketing and technology strategy.

In Prediction Machines, Agrawal, Gans, and Goldfarb argue that artificial intelligence is best understood as a technology that makes prediction cheaper. AI helps businesses use data to make better guesses about what will happen next, what something means, or what action is likely to work.

The core thesis is that when prediction becomes cheaper, companies need to rethink decisions, workflows, jobs, and strategy. AI does not remove the need for human judgment. It makes judgment more important because leaders still have to decide what outcomes matter.

AI is not magic. It is a prediction tool.

It can help answer questions like: Which customer is likely to leave. Which loan is likely to default. Which applicant may perform well. Which product will a shopper probably buy. Which machine might fail soon. Those predictions can be very useful, but they are not the same as decisions.

This matters because businesses often rush into AI without knowing what decision they are trying to improve. The smarter move is to break work into parts: prediction, judgment, action, and outcome. AI can improve the prediction, but humans still need to choose the goal, weigh tradeoffs, manage risk, and decide what “better” actually means.

If this idea resonated with you, share it with your network using the social sharing buttons at the top of this post.

Start by listing the important decisions your business makes often. Look for decisions in sales, hiring, pricing, customer support, operations, inventory, marketing, lending, risk, or product development. Then ask which parts of those decisions depend on prediction. If a team is trying to guess demand, churn, quality, fraud, fit, timing, or intent, AI may be useful.

Next, separate prediction from judgment. Do not ask AI to “make the decision” unless you are clear about the goal and the tradeoffs. A model may predict which lead is most likely to buy, but a leader still needs to decide whether the company wants fast sales, better-fit customers, higher margins, or long-term retention. Those goals can point to different actions.

Then look at your data. AI depends on useful information. If your data is messy, incomplete, biased, or trapped in disconnected systems, the predictions may be weak. Before chasing advanced tools, make sure the business is collecting the right data, cleaning it, and connecting it to real decisions.

After that, redesign the workflow around the prediction. If AI can flag a customer at risk of leaving, decide who sees that flag, what action they take, how quickly they act, and how success is measured. A good prediction that does not change the process is just an interesting report.

Finally, start with a narrow use case. Pick one decision where better prediction would clearly improve cost, speed, quality, or customer experience. Test it, measure it, and learn from it before spreading AI across the company. The goal is not to look advanced. The goal is to make better decisions.

Share the insights below on LinkedIn or X/Twitter and we’ll feature your business in the newsletter. Just use the hashtag #BizBookDaily. It’s as simple as that.

Ayush Karekar — SEO-Driven Content Writer at InAmigos Foundation (IAF) — Follow him on LinkedIn if you’re looking for SEO-friendly content writing and discoverable digital content.

We’ve been collecting standout business insights from experienced operators—short, practical ideas that hold up in the real world. Take a look at our Top Insights here.

Agrawal, Gans, and Goldfarb provide a whole lot more useful info in Prediction Machines. Here are three reasons you might want to read the full book:

Consider skipping this book if you want a technical guide to building machine learning models.

Sponsored by

Many of the best business books begin as ideas leaders have been refining for years—frameworks, philosophies, or lessons learned the hard way. Writing the book forces those ideas into clear form, and once they’re on the page, they can travel far beyond a single conversation or presentation.

At MemoirGhostwriting.com we write memoirs and business books for entrepreneurs, founders, executives, and industry experts. Most of our clients have been meaning to write their book for years—they just don’t have the time to sit down and do it properly. We help them speak more frequently on stages, scale consulting practices, attract more leads and move them deeper down their marketing funnels, increase conversion rates in complex sales cycles, and position them for for-profit boards.

We don’t just help you write, we take the whole project off your plate and see it through to a finished, publishable book.

Our team handles the entire process:

If you’ve ever thought about what a best-selling business book could do for you, let's talk.

Learn more:

Know someone who likes smart business ideas?

Forward this email to one colleague or friend who would enjoy today’s lessons.

Or send them here:
BusinessBookDaily.com