Alexi Horowitz-Ghazi

At its core, Stephen is trying to estimate how many copies a book might sell.

To do that, he starts by building a model.

At first, whenever a new book idea is being seriously considered for acquisition, the model is just based on the comparable titles and how they've sold over time.

This is what helps Norton determine whether they buy the book and for how much.

Once Norton is committed to publishing a book, and as it moves through the production pipeline, Stephen's job is to keep updating the model.

To do this, he's constantly running a kind of intelligence operation, hoovering up the market signals from his vast sales force.

Norton has sales reps for the independent bookstores in each part of the country, and there are reps for bookstore chains and online retailers, all negotiating over how many copies of a given book they will order.

The orders are the intelligence that comes back to us.

Because as Norton's sales reps are hitting the ground, pushing the Planet Money book to the thousands of book buyers like Fisher all around the nation, they are also running those order estimates back up the chain to Stephen in Central Command.

He's then feeding all that information into his model to decide the number of copies of a book Norton will print in the first run.

He needs to print enough books to supply all the estimated demand.

Stephen's challenge is kind of a version of what Fisher faces at Carmichael's bookstore, but at a much bigger scale.

He needs to ensure there will never be a customer in any bookstore asking for a Planet Money book and coming away empty-handed.

So he doesn't want to print too few copies, and he wants to be able to print later runs of the book in a quick and convenient way.

By, for example, printing the book domestically.

At the same time, if he prints too many copies, he'll risk clogging up Norton's warehouse.

This is what we call the opportunity cost, right?

Right.

The more time the books sit there unsold in the warehouse, the more they cost, by crowding out books that might sell faster.

One of the biggest potential costs in figuring out the print run is the number of returns that might be coming back.