The bottleneck

Why most teams cap out at one video a month

Three quarters of companies publish at least one video a month and almost none publish more. The ceiling is production capacity, not ideas — and the teams that broke through it changed the workflow, not the budget.

The ceiling is real and it is low

Wistia's 2026 report surveyed over 900 professionals and analysed more than 13 million videos on its platform. Seventy-six percent of companies publish at least one video a month. That sounds like health until you notice where the distribution stops: the teams publishing at real volume are a narrow band, and the two barriers cited most often are company size and resources, and cost.

Set that against the budget picture. Only around 40% plan to increase video spend into 2026; 46% are holding flat. So the demand for video is rising while the money for it is not. Something has to give, and historically what gives is output — a team decides it can sustain one video a month and builds its calendar around that number.

Cost per video is the wrong thing to optimise

The instinct is to make each video cheaper. Cheaper stock, shorter edits, a junior editor. That lowers the unit cost and leaves the actual constraint untouched, because the constraint is not money per video — it is the number of hand-offs each video has to survive. Brief to script, script to voice, voice to edit, edit to review, review to export. Each hand-off is a queue, and queues are where calendars go to die.

The teams that broke the ceiling did not find more budget. They removed hand-offs. Wistia's data shows the correlation clearly: teams using AI somewhere in the workflow are 57% more likely to publish 50–100 videos a year, and twice as likely to publish 100–250. Worth reading that honestly — it is a correlation from a self-reported survey, not proof that a tool caused the output. Teams that adopt AI early are probably already unusual. But the direction is consistent, and it matches the mechanism: fewer queues, more throughput.

What that looks like in practice

Orb Studio's Agent mode collapses the whole chain into one scheduled job. It researches your brief, writes a roughly 30-second script, voices it, art-directs the scenes against a library of about 6,250 line-art icons, choreographs the motion, and renders a finished MP4 — while the app is open and without anyone in the loop.

The part that matters for sustained output is the queue. The agent plans a slate of upcoming clips ahead of time rather than improvising one at a time, and a uniqueness engine fingerprints hooks, staging and choreography so it never ships the same video twice. That is the difference between a tool that makes a video and a workflow that makes videos.