Answer in brief
A platform dependency is acceptable when ownership, access, export, support, security updates and exit conditions are explicit. The risk is not using another company’s platform; it is discovering the dependency after launch.
The field that does not exist
A VIT MARKET listing publishes thirteen fields. A number, a title, a difficulty word, a category, a description, two price strings, a delivery estimate, a slug, a group key, a price floor, a price ceiling and a featured flag. Not one of them names a prerequisite, an account, a subscription or a platform. There is no field that says what this work will run on when it is finished.
That is not unusual — most service catalogues are built the same way — which is why the question has to be asked before you commission. The price tells you what you owe the contractor. It is silent about the account you will hold and the second bill that arrives from someone you have never spoken to.
So VJOURNAL read the only fields where a platform can appear at all: the title and the description of each of the 100 listings, in all three language editions the site serves. A name was counted only where the Russian, English and Spanish text of the same listing all carry it, so that nothing in the count is an artefact of one translation. The pass returns 35 listings, 34 distinct third parties and 57 mentions.
Thirty-five listings, thirty-four names
Sixty-five listings out of a hundred name nobody. Of the 35 that do, 21 name one platform, nine name two, two name three and three name four. The three that name four are the GEO and AEO listing — Google, Yandex, ChatGPT, Perplexity; the social-media autopilot — Telegram, Google, VKontakte, Zen; and the review collector — Telegram, Yandex, 2GIS, Otzovik.
The full roll of names, ordered by how often each appears: Telegram eleven times; Google nine; Yandex three; Ozon, Shopify and Wildberries twice each; and once each Astro, Next.js, WordPress, Tilda, Wix, Figma, Notion, Behance, ThemeForest, YouTube, ChatGPT, Perplexity, Yandex Alice, n8n, Make, Albato, amoCRM, Bitrix24, Outlook, Gmail, Apple Mail, Slack, Supabase, Excel, 1C, VKontakte, Zen, 2GIS and Otzovik. Twenty-eight of the 34 appear exactly once in the whole catalogue.
One boundary changes the number, so it belongs here. Schema.org appears in a listing title and was excluded: it is a shared vocabulary, not a platform anyone holds an account on. Add it back and the count reads 36 listings and 35 names. Everything else on the list has an owner who is neither you nor your contractor.
A platform name does three different jobs
Not every name is a commitment, and treating them alike would produce a nonsense shortlist. Reading the 57 mentions one at a time, three roles fall out. Fifty of them are dependencies: the delivered thing runs on that platform, writes into it, or is measured against it. Three are exits — the listing sells getting you off the platform it names, and all three sit in two listings, one for WordPress and one covering Tilda and Wix together. Four are pure contrast: the platform is named only to say what you are not getting.
The contrast cases repay a close reading, because at a glance they look like dependencies. The landing-page listing promises a working site on your own domain “rather than a Figma mockup”. The portfolio listing offers a personal site instead of a link to Behance, the documentation listing offers real docs instead of scattered pages in Notion, and the multilingual listing promises search traffic abroad rather than a Google translation. Four names, zero commitments.
Sorted by listing rather than by mention, 30 listings name at least one platform the delivered work will actually depend on. Two name a platform only as something you are leaving. Three name one only as a rhetorical contrast. Sixty-five name nothing at all, which is a different situation again: there, the stack is undeclared, and the choice — with whatever recurring cost it carries — is still open at the moment you read the price.
Telegram is named more often than Google
The single most-named third party in this catalogue is Telegram, at eleven listings. Google follows at nine. Nothing else reaches four. That ordering is a fact about what the shop actually sells: alerts, bots, report deliveries and internal assistants all terminate in a chat, and in this catalogue the chat is usually Telegram.
It also has a consequence a price cannot show. A bot needs a token, a channel needs an owner, and both belong to an account. If eleven of the jobs you might commission end inside one messaging platform, who registers that account — and who still controls it after the final invoice — is the difference between owning the automation and renting it.
The named group is also the cheaper group, which is the opposite of what most people expect. Taking the midpoint of each published band, the 35 listings that name a platform have a median midpoint of $525 against $800 for the 65 that name none. Their means are almost identical — $1,014.57 against $992.31 — so the difference is shape, not size: naming a platform clusters at the cheap end, with a few large exceptions pulling the average back up.
The dearest shelf names the fewest platforms
Split the 30 dependency listings by group and the distribution is lopsided: 13 on automation, nine on web development, five on search and content, three on the AI shelf. Every group holds exactly 25 listings, so the AI shelf declares an outside platform three times out of 25 — the lowest share in the catalogue, and the shelf whose work most obviously runs on somebody else's model.
The catalogue does acknowledge this, but not in any individual listing. The AI group's shared deliverables bullet — the one line that changes with the group — promises response-quality evaluation, cost limits, logging and a safe shutdown procedure. Cost limits: a phrase on all 25 AI pages and nowhere else, and the only structural admission in the catalogue that this work carries a bill that keeps arriving.
One listing goes further, and it is the exception that proves the point: the prompt audit, published at $800 – $2,500, whose entire subject is reducing an API bill by two to five times. It is the only listing in the catalogue whose product is a running platform cost rather than a project. It is also, necessarily, a bill the catalogue never quotes — because it is yours already.
Someone else's bill appears five times, always as money you stop paying
Read all 100 descriptions for any mention of a recurring third-party cost and five come back. The Shopify store arrives without hosting. The WordPress migration does not require paid hosting. The spreadsheet-driven catalogue site removes the need to pay for a CMS, and the MDX blog says the same about hosting and an admin panel. The builder migration drops a monthly subscription and claims the saving pays for the work in six to twelve months.
All five point the same way. In every case the platform fee is presented as money the buyer stops spending. Not one of the 100 descriptions quotes a platform fee the buyer starts spending, even where the listing sells a build onto a platform that is not free to keep.
This is not a trick — a description written to sell a website may lead with what improves. It is a structural gap the buyer has to close, because the two numbers on the card are a project price, and the number that decides whether the project is affordable in year two is not on the card at all.
The exits it prices, and the ones it does not
Line the entrances up against the exits and an asymmetry appears. The catalogue prices a way off WordPress and a way off Tilda and Wix, and no way onto any of the three. In the other direction it prices two ways onto Shopify — a turnkey store and a custom theme — plus a landing page on Astro, a corporate site on Next.js, a customer area on Supabase and a template for sale on ThemeForest. For those six there is no exit anywhere in the hundred.
That is a fair reflection of what a shop believes rather than a hidden defect: leaving a website builder is a job people pay for, and leaving a modern framework usually is not. But a buyer should notice which side of that line each listing sits on. A migration listing is priced evidence that the platform it names has an exit cost — and that somebody once paid it.
The reverse reading is the useful one. Before commissioning anything from the 30 dependency listings, ask what the equivalent exit would cost, and notice that the catalogue answers that question for three platforms and stays silent about the rest.
What a platform name cannot tell you
A name in a description is not a specification. It does not say which plan, which region, which limits, or whose name the account is in. Nor does silence mean independence: 65 listings name nobody, and several of them almost certainly touch a platform anyway — a hosting provider, a mail relay, a model endpoint. The data supports only that those listings do not say so.
The classification into dependency, exit and contrast is VJOURNAL's reading of 57 sentences, not a field in the dataset. The counts are reproducible; the labels are a judgement, published listing by listing so that anyone who reads a sentence differently can move it and recount. The only mechanical guarantee in the method is the three-edition rule, which removes names that exist in one translation and not the others.
Nothing here supports a claim about what any named platform charges, and no such figure appears in this article. The catalogue publishes no third-party prices, VJOURNAL fetched none, and a comparison quoting them from memory would be worth less than the silence it replaced.
Four questions before you commission
First: name every outside platform this work will touch, including the ones the listing does not mention, and write them into the brief. If the listing names one, that half of the answer is free; if it names none, ask out loud. Second: whose account, whose card, whose recovery email? Ownership of an automation is ownership of the account it lives in.
Third: what does that platform cost per month at the volume you actually expect, and who is on the hook when the volume doubles? The catalogue's AI shelf calls this a cost limit and puts it on all 25 of its pages; treat that as the template for the question, not as an answer to it. Fourth: what would it cost to leave? For WordPress, Tilda and Wix this catalogue publishes a number. For everything else, the number exists but nobody has quoted it to you.
None of this makes a platform-named listing a worse buy than a silent one. If anything the reverse: a listing that names Telegram, Shopify or n8n has told you something real about what you are signing up for. The point is that the naming happens in a sentence, not in a field — and a sentence is only load-bearing once you have copied it into the brief.
Decision framework: platform ownership in a web project
A platform dependency is acceptable when ownership, access, export, support, security updates and exit conditions are explicit. The risk is not using another company’s platform; it is discovering the dependency after launch.
CISA acquisition guidance asks suppliers to disclose third-party components, provenance, support and end-of-life practices. Those questions apply directly to builders, plugins, hosting and external APIs.
Create a dependency register with account owner, billing owner, data location, export format, renewal date, replacement path and incident contact for every external service in the proposal.
Practical checklist
- platform ownership in a web project: write the promised outcome and acceptance rule.
- platform ownership in a web project: record every exclusion, dependency and unresolved assumption.
- platform ownership in a web project: assign an owner and review date to evidence that can narrow the estimate.
- Create a dependency register with account owner, billing owner, data location, export format, renewal date, replacement path and incident contact for every external service in the proposal.
Questions and answers
platform ownership in a web project: what should a buyer verify first?
A platform dependency is acceptable when ownership, access, export, support, security updates and exit conditions are explicit. The risk is not using another company’s platform; it is discovering the dependency after launch. Create a dependency register with account owner, billing owner, data location, export format, renewal date, replacement path and incident contact for every external service in the proposal.
platform ownership in a web project: which assumption changes the estimate most?
CISA acquisition guidance asks suppliers to disclose third-party components, provenance, support and end-of-life practices. Those questions apply directly to builders, plugins, hosting and external APIs.
platform ownership in a web project: what is the next practical step?
Create a dependency register with account owner, billing owner, data location, export format, renewal date, replacement path and incident contact for every external service in the proposal.