The Pivot to Video Didn't Fail Because of Video

Every publisher who lived through 2016 filed the same incident report: we pivoted to video, video didn’t work, don’t pivot to video. Ten years later that report is still getting cited as gospel, and it’s wrong on the one line that matters. The postmortem named the wrong cause of death.

TL;DR: The “pivot to video” collapse wasn’t a video problem. It was a subsidy problem. Publishers built newsroom capacity against Facebook’s algorithmic boost and ad payments, and when Facebook pulled the subsidy, the capacity had no revenue under it. Simon Owens argues that outlets citing that history to dismiss the New York Times’s current video expansion are pattern-matching on the surface feature (video) instead of the mechanism (who’s funding the build-out and why). Get the mechanism wrong and you’ll avoid the right moves for the wrong reasons, which is its own kind of expensive.

The format did not kill the first pivot to video

Not the format. The dependency.

I build a log for a living, so let me put it in the only terms I trust. A schedule is a promise and every position on it is sold to somebody.

Those newsrooms did not build a video capacity. They built positions against an advertiser who was not on the contract and had never signed anything.

When Facebook stopped buying, the log did not change. The as-run did. That gap is the whole story, and nobody read it that way at the time. Including me.

Facebook spent several years in the mid-2010s algorithmically boosting video in the feed and, for a stretch, paying publishers directly to produce it. That looked like a market signal. It was closer to a grant program with an undisclosed sunset date. Newsrooms built full video teams (producers, editors, on-camera talent, the works) sized to the traffic and the checks Facebook was issuing, not to what the content could earn on its own. When Facebook decided video no longer served its feed strategy and throttled both the payments and the distribution, every newsroom carrying that headcount was suddenly running a cost center with no revenue attached. Mass layoffs followed. The trade press wrote it up as “video doesn’t work for publishers.”

That’s a diagnosis error you’d flag in any ops review. The variable that failed wasn’t the content format. It was capacity built against a subsidy the publisher didn’t control and couldn’t audit. You could swap “video” for “SEO traffic,” “newsletter referrals from an aggregator,” or any other externally-granted distribution channel and get the identical failure mode. The format is a red herring. The dependency is the story.

Why does the Times’s video bet look the same and run different?

Because everyone’s checking the wrong column.

Owens points out that the Times is expanding video investment right now, and the knee-jerk response from a chunk of the industry is “here we go again.” But run the mechanism check instead of the pattern match. The Times funds this expansion out of a profitable subscription business it controls end to end, not out of platform bonus payments it can lose in a policy meeting it wasn’t invited to. The distribution ecosystem for video has also matured past 2016’s Facebook monoculture: YouTube, Instagram, and TikTok now recommend on engagement and merit signals rather than a single platform’s arbitrary algorithmic favor of the moment, and creators on those platforms generate real, durable revenue at scale. Same output format as 2016. Completely different capital structure underneath it.

This is the audit you’d run on any recurring commitment before you sign off on it: who’s funding the build, what happens to the build if the funder walks, and does the team doing the work control the channel or just rent access to it. Two outlets can run the identical playbook and get opposite outcomes because the ledger underneath the playbook is different. A postmortem that doesn’t ask that question is a caption.

What does a football team have to do with your publishing calendar?

Same mechanism, different building.

Owens’s other thread this week is sports organizations quietly becoming media companies. His example is the Kansas City Chiefs producing their own video content, selling sponsorship around it, and hiring outside creators to host shows that aren’t game footage at all. That’s a franchise doing on purpose what the 2016 newsrooms did by accident: building direct-to-audience distribution it owns instead of renting attention from a platform’s news feed. The team isn’t chasing a viral moment. It’s building a channel it controls, funded by revenue it already has, aimed at an audience it already owns. That’s the healthy version of the same move the collapsed newsrooms made in the unhealthy direction.

If you’re running any kind of publishing or content operation, this is the register that question belongs in, not “should we do more video.” Ask instead: which parts of our distribution do we own, which parts are we renting, and if the rent went to zero tomorrow, what would still be standing. The answer to that question tells you more about your actual risk than any format decision does.

Frequently asked questions

Does this mean video is always a safe bet for publishers now?

No, and that’s not the claim being made here. The claim is narrower: don’t reject video (or any format) on the strength of a decade-old failure without checking whether the failure mode that killed the last attempt is actually present in this one. A subscription-funded, owned-audience video build carries a different risk profile than a platform-subsidized one. Both can still fail for other reasons: bad content, wrong audience fit, poor execution. The mechanism check rules out one specific historical trap. It doesn’t rule out the rest of the job.

How do I check whether my own project has this dependency risk?

Trace every recurring cost back to its funding source and ask what happens to that cost if the funding source disappears with no warning. If the answer is “the team gets cut” or “the project stops,” you’re running on a subsidy, not a business, whether the subsidy is a platform’s algorithm, a single client, a grant, or an aggregator’s referral traffic. That’s not automatically wrong to build on. It’s wrong to build on without knowing you’re doing it.