I want to open this essay with a specific number, because the number was, in the end, what made me write the essay.
The number is 137. In late 2024 I sat down with a friend who runs the video function at a mid-market UK apparel brand — call her Julienne — and we counted how many pieces of short-form video her team had produced in the previous 90 days across TikTok, Instagram Reels, YouTube Shorts, and the brand's own site. The answer was 137. The team was six people. The fully-loaded cost of the programme, on Julienne's honest calculation, was roughly £220,000 for the quarter, or about £1,600 per video.
The revenue attributable to the programme, on the brand's own attribution model — which was, by industry standards, quite generous — was approximately £310,000 for the quarter. On the model, the programme was net positive by a small margin. On Julienne's private MMM-triangulated view, the programme was somewhere between break-even and modestly negative. On any read of the numbers, the programme was not producing the sort of return that had, four years earlier, justified investing in short-form video as a category.
Julienne, when I asked her whether this pattern was unique to her account, thought for a while and said: "Everyone I know is in roughly the same position. Nobody talks about it because the industry hasn't given us permission yet." This essay is an attempt to give the permission.
What changed
The five-year story of short-form video, in commercial marketing, breaks roughly into three phases.
The first phase, from roughly 2020 through the middle of 2022, was one of straightforward opportunity. The platforms — TikTok primarily, but Instagram Reels and YouTube Shorts once they existed — were aggressively distributing short-form content to broad audiences. The audiences were young, receptive, and demonstrably converting on strong short-video creative. Per-video economics were exceptional. Brands that entered the category early and shipped consistently produced remarkable returns.
The second phase, from mid-2022 through late 2024, was one of professionalisation. As the platform algorithms matured and the audiences became more sophisticated, the amateur short-video that had defined phase one stopped performing. Production standards rose. Editorial standards rose. The teams shipping short video became larger and more specialist. Per-video cost rose commensurately. Per-video return, on our data, held roughly stable — the improved production quality justified the higher production cost.
The third phase, which I would date to roughly the first half of 2025, is the one this essay is really about. In this phase, the platform algorithms have continued to raise the bar on what surfaces to broad audiences, while simultaneously the audience-side saturation of short-form commercial video has reduced the audience's per-video receptivity. Production quality still matters — probably matters more than ever — but it has stopped being sufficient. The per-video return, on our data across 91 brands, has fallen substantially from the phase-two baseline. The per-video cost, meanwhile, has held or continued to rise.
The result, on the arithmetic, is that most brand short-video programmes in 2026 are running at meaningfully worse unit economics than they were in 2023. The industry's default advice to ship more videos, iterate faster, produce more variations was, in phase two, defensible. In phase three, it is producing programmes that lose money per video shipped, on the accounts we audit.
What the data shows
The dataset I have been working from covers 91 brands across DTC, retail, subscription, and B2B SaaS, all UK or European, all producing short-form video across at least two of TikTok, Instagram Reels, and YouTube Shorts. Combined quarterly short-video output across the sample runs to somewhere north of 8,000 pieces per quarter. All the brands agreed to share aggregate attribution data and to be included in an anonymised analysis.
The headline finding is that the median brand's revenue-per-video, measured on the brand's own best-available attribution model, fell approximately 34% between Q1 2024 and Q1 2026. The interquartile range of this decline was wide — from -12% to -58% — but the direction was almost universal. Only 11 of the 91 brands in the sample showed year-over-year improvement in revenue-per-video across the window.
The mechanism, on inspection, was consistent. Short-video output volume, per brand, rose approximately 22% over the same window. Median reach per video fell approximately 41%. Median engagement rate per delivered impression fell approximately 26%. The platforms distributed less, the audience engaged less with what they saw, and the attributable downstream conversion consequently declined per video shipped, even as more videos were being shipped.
The brands that showed year-over-year improvement, when we looked at them specifically, had something in common. They had, on average, reduced short-video output volume by roughly 30% across the window, and had substantially increased the editorial investment per video. Their per-video revenue rose because their per-video cost rose more slowly than the industry median, and their per-video output declined enough to increase the attention per piece. The compensating dynamic was clear.
"The path that produced the most improvement in short-video economics, on our sample, is the exact opposite of the path the industry has been recommending. The industry's advice was more videos, faster. The brands whose economics improved shipped fewer videos, better."
What to do
If you are running a short-form video function right now, three practical implications, in decreasing order of impact.
The first is to audit your per-video economics. Pull twelve months of production output. For each video, note the production cost (fully-loaded, including team time, creative agency fees, and platform amplification if any). Estimate the attributable revenue on your best-available attribution model. Sort by revenue-per-cost ratio. Look at the distribution.
The distribution will, on our sample, almost certainly show a long tail of videos where the per-video ratio is negative. The interesting question is what share of the total output is in that tail. On the median brand in our sample, roughly 40-55% of the total videos produced were in the negative tail. Removing those videos from the calendar — not replacing them, removing them — is the single most reliably productive intervention available.
The second is to raise the editorial bar on what does ship. The brands that improved their per-video economics did so by shipping fewer videos with meaningfully more editorial investment per video. This is, in practice, harder than it sounds. The team's habits are calibrated to a specific cadence and a specific per-video effort level. Shifting to a lower cadence at a higher per-video effort level requires the team to relearn its own workflow, and produces uncomfortable delivery-count reductions on internal reporting for the first two to three months.
The third is to have the honest conversation with your CMO about what short video is actually contributing. Most short-video programmes are, on inspection, being reported using metrics — impressions, views, engagement counts — that are structurally decoupled from the attributable revenue the programme is producing. The CFO's view of the programme, based on the finance model, is often materially less flattering than the CMO's view, based on the marketing report. Aligning the two views produces internal conversations that are politically difficult but commercially necessary.
The hard part
The hard part, in my read of the accounts that have made this transition successfully, is not the analytical work. The analytical work is straightforward and mostly conclusive. The hard part is the operational and cultural work of accepting that a category the industry has universally endorsed for five years has entered a different phase of its economic life, and that the programmes built for the earlier phase need to be substantially reshaped.
The teams that have made this transition on the accounts I have watched are, in every case, smaller than they were during phase two. They ship less. They earn more. They spend more time per piece and less time on the meta-work of coordinating output. Their weekly reporting looks emptier. Their finance reporting looks stronger. Their CMO relationships, once the transition is complete, are more constructive than they were during the phase-two output-heavy period.
None of this is comfortable to recommend. Recommending it involves telling teams that have grown their reputation on short-video output that the output is now, in aggregate, working against them. It involves telling agencies and vendors that the short-video packages they have sold successfully for four years are, at 2026 market conditions, over-producing at the wrong end of the value curve. It involves telling CMOs that the reporting model they have been running is producing a systematically flattering view of a programme that has, on the underlying economics, quietly turned.
The reckoning is happening whether the industry writes about it or not. The brands that see it early and adjust will hold a substantial advantage over the brands that continue to ship more, faster, on the older advice. The industry will, in the end, follow the advice the current advantage produces. Better to be the source of the new advice than the last person following the old one.