The audience KPIs that might have saved WaPo's Book World section

When news broke last month of the layoffs at the Washington Post, the narrative painted owner Jeff Bezos as the villain. After reading former Book World editor John Williams' op-ed in The Atlantic, I wonder if that’s too easy.

Because with a deck that reads, "The editor of the recently scrapped Book World believes in serving subscribers, not data," I think the matter deserves some unpacking.

Williams’ op-ed reads like an editor who knows times have changed, but refuses to believe it. Bezos’ philosophy, as Williams describes it, is essentially that data should inform editorial strategy, and Williams even acknowledges that looking at what readers click is “something any sane 21st-century journalist does.” And then he completely misses the point.

“As a reader of many distinctive publications, I want to be led by them,” he writes. “What makes them special is where they choose to take me, and how much I trust them to do that.”

This made me mad. I’m a Washington Post subscriber. The layoff news hit me hard. I sent my favorite Post writer an "I'm-glad-you're-still-here" email because I love his work. I was laid off late last year, so I feel for Williams. But the fact that someone in the newsroom cared more about leading readers than listening to what audiences wanted made me furious. This isn’t a Bezos problem (not entirely anyway), and this didn’t have to happen.

So here's how I would've coached Williams to think about data if I was his audience editor. Maybe he and the audience team did everything I'm about to suggest. By writing this, I hope to offer a simple starting point for editors thinking about audience data in journalism.

First, I would have narrowed the key performance indicators that he should focus on to new subscriptions, total traffic and subscriber traffic. For editors, those three will tell you a lot about what readers want from you.

An article that hits the gold standard—high new subscriptions, traffic and subscriber traffic—means it interested current readers and brought in new subscribers. High subscriptions but low subscriber traffic means there may be a new audience worth pursuing. Low subscriptions but high subscriber traffic means you're pleasing your current audience. For traffic in general, use the average amount of traffic over the last three or six months. This baseline will help you determine where the story landed with readers.

Then we’d look at the top-performing stories and look for patterns in topics or article types. What topics brought people in? Did they read book reviews? If so, which books? Then, what styles resonated with readers? And finally, what else about those stories might have been appealing? Were they short or long? Was the headline punchy or straightforward?

This process also works in reverse. It hurts to see good stories flop, but that data is the audience telling you exactly what they don't like. Our news judgment shouldn’t be the best or only audience barometer any more.

From these three KPIs, Williams could’ve built an editorial strategy centered on the topics and article styles that worked well with new and current readers. When this strategy is working, you don't have to "lead" readers. You know what they like. You know what they want from you. You can take a "leading" topic and spin it in a way that still connects.

All of this could be bolstered by an audience team A/B testing headlines and coordinating traffic distribution. I hope I’ve just described something that Williams did do. That way, we can all blame Bezos and go home. But if it didn’t, then it might be time to consider that leading is not an audience strategy in the digital age. Maybe it never was.