Optimization won't fix local news

Local news organizations cannot optimize their way to relevance.

Or such is my biggest takeaway from a new report on what kinds of local news stories are most likely to turn readers into subscribers.

The good news, for those of us who still believe in the civic mission of journalism, is that hard news matters. The bad news is that even those stories do not convert enough readers to pay for the work required to produce them.

That’s the part we need to sit with.

Especially over the last decade, much of the local news business has been built around the idea that we can optimize our way to sustainability. If we A/B test our headlines, and tune our churn models, and make more decisions based on analytics, all that will add up to more subscribers and ultimately sustainable revenue.

Most of those things have been necessary and long overdue. But this report is yet another reminder that they aren’t sufficient. Not even close.

Continuing to do what we are currently doing — only 20 percent more, faster, better, or more efficiently — will not reverse the fundamental deficit in relevance that local news has accrued over many decades.

That’s one reason why I remain skeptical of AI products and strategies that focus primarily on output and efficiency.

Large language models are powerful tools that can unlock new ways of telling stories and doing business. But to be effective, they need to be coupled with coherent strategies that create new value for readers and advertisers.

News organizations have more opportunities than ever to do that, if they’re willing to make the investments and absorb the risk. (I would count the Minnesota Star Tribune’s push into prep sports as one example.)

Otherwise, AI becomes just another optimization layer on top of a product too many people have already decided they can live without.