Disclaimer first: this company does not exist. It's a composite — call it a 12-person video production company that ships client work and internal content every week. Every subscription price below is a real market price you can check yourself. The point is the math, and the math is generic: replace "video" with your industry and the shape holds.
What they rent today
Four subscriptions, three humans doing unpaid tool-work between them:
-
01
Video review & file sharing
A Frame.io-style platform so clients can timecode-comment on cuts instead of writing "the part at 2:14" in email. ~€19/seat/mo × 3 seats ≈ €57/mo. It works. It's also rented forever.
-
02
File delivery & handoff
An editor exports a master, uploads it to a shared folder, then… tells someone it's there. By Slack message. Every time. The upload part is subscriptioned; the "hey, it's up" part is a human doing integration work for free. ≈ €15/mo plus maybe 40 minutes/week of someone's salary.
-
03
Premiere Pro plugins, bought piecemeal
Auto-subtitling, upscaling, format conversion — each a small subscription or per-export fee scattered across vendor accounts. Nobody in the company knows the real total; finance estimates €46/mo when asked.
-
04
The photo pipeline, done by hand
New product shots arrive, someone manually crops them to spec, renames them, drops them in the right folder, pings the web person to upload. A recurring ~€150/mo of someone's time disguised as "just part of the job." The cropping itself takes minutes; the notifying, renaming, and waiting take hours.
Itemized: €57 + €15 + €46 + €150 ≈ €268/month, or €3,216/year — plus the tax nobody line-items: three skilled people spending hours a month being middleware.
What an owned pipeline costs
Same four jobs, built once, owned forever:
-
A
Review & approval page
A private page where editors drop a cut and clients timecode-comment. No seat pricing — everyone in the company and every client uses it. Built on the studio's flat price sheet as a €690 full site (it's a website with auth and comments, not magic).
-
B
File watcher with notifications
A watcher on the delivery folder: file lands → relevant people get notified with a link. The thing the €15/mo tool does, minus the part a human does anyway. €350 automation build.
-
C
Pipeline glue for the plugins
Most Premiere plugin jobs are scriptable outside the editor — subtitle generation, upscaling, format conversion run as batch jobs on watch folders. Some stay as one-time plugin purchases; the recurring glue is €350 automation build + pennies of compute.
-
D
Photo pipeline, automated
New photos land in a folder → auto-cropped to spec, renamed to convention, sorted, and the upload person gets a notification with everything ready to go. This is the €150/mo of human middleware, built as a €350 automation.
Total owned: €1,850 one-off. Hosting for the review page: ~€5/mo.
The deeper cut: the production line itself
The tools above are the visible costs. The bigger one is upstream — how a video gets made today at this hypothetical shop:
-
05
Somebody writes the script
Hours of drafting, restructuring, rewriting the intro. The research and the talking points are human work; the first and third drafts don't have to be.
-
06
Somebody fact-checks it
Claim by claim, usually by opening twenty browser tabs. Slow, important, and mostly mechanical: find the source, confirm the number, note the citation.
-
07
Somebody breaks it into scenes
Script → beat sheet: what's on screen for each line, how long each shot holds. Structured work that follows rules a machine can follow too.
-
08
Somebody routes every scene
Which scenes need 3D animation? Which are stock footage? Which could be AI-generated? And then the actual hunting: finding stock clips, writing image prompts, generating takes, naming files, moving them where the editor expects them.
Here's the claim, and it's the one that matters: 80–90% of steps 05–08 is automation work. Draft from the brief with the company's tone and past scripts in context. Fact-check as retrieval with citations attached — every claim linked to its source, so the human check takes minutes, not evenings. Beat sheets parsed straight into a structured scene list. Scene routing run as classification against the company's own rules ("talking-head → stock, product-on-table → 3D, b-roll of concept X → AI gen, per the style guide"). Then the machine does the fetching, prompting, generating, and filing.
What stays human, deliberately: the quality control pass. One person reviews the machine's output — script tone, fact-check spot samples, whether the generated footage actually matches the scene intent — and approves or bounces it back. The editor still makes the film. The pipeline just stops making the editor do clerical work.
The honest math on this one: it's not a €268/mo subscription, it's salary — a junior researcher/PA role at maybe €2,000/mo doing work that's mostly mechanical. Automating the 80% and keeping a human QC loop doesn't eliminate the role, it changes what the role does and how many videos the same team can ship. That's a conversation with numbers attached, not a flat price in a table — which is exactly why it gets its own brief instead of a line item here.
The crossover month
€268/mo rented vs €1,850 built once. The owned pipeline pays for itself in month 7. From month 8 on, the company saves ~€263/mo — €3,156/year — and the tools keep working after that, at no increase, forever. If a subscription price rises 10% (they do), the crossover moves earlier and the annual savings grow.
When does renting win? If the company dissolves within 6 months, or needs enterprise compliance certs day one, or the tool sits at the center of a network nobody can recreate. None of those describe a 12-person shop with a folder of files.
The honest caveats
The rented tools come polished and supported; the owned ones come with an owner. Someone has to click "deploy update" when a codec changes, and the €1,850 assumes one round of scope — add features, add cost. A rented tool hides those costs in the price; an owned one shows them to you. This studio's bet: shown beats hidden.