A new site takes weeks to be found. Here is how Jimp does it in an hour.

A new adult site is invisible for weeks, and not because it is bad. It is invisible because nobody told anyone it exists. The search engines will find it eventually, when a crawler with nothing better to do follows a link that leads there, and that "eventually" is the part you pay for in silence. We spent the last weeks removing it from Jimp sites. Every video a site publishes is now named to the engines within the hour, posted on the site's own Telegram channel and Bluesky account, and drawn as a share card wherever its link is pasted. Nothing to click. Here is what that actually does, and what it does not.

What a crawler does with a domain it has never heard of

A crawler budgets its visits. Google documents it as crawl budget: a site gets the number of fetches its servers and its perceived value can bear, and the demand side comes from how popular and how fresh its addresses look. A domain registered last month, with no links and no history, starts at the bottom of that list. It gets a first visit when something points at it, and then a few pages at a time, and then a return visit when the first pages turned out to be worth it.

That arithmetic is fine for a site that already has an audience. It is brutal for a site that has none, because the only way to earn visits is to have been visited. We wrote about the other half of this problem earlier, the launch flood that buries a new site under pages it will never get indexed. This one is the flip side: even at the right pace, a page nobody announced waits.

Tell them, do not wait for them

The fix is old and most sites skip it. The engines accept to be told. Bing, Yandex, Seznam and Naver share one protocol for it, IndexNow, where a site posts the addresses that changed and a key it serves itself to prove the addresses are its own. Google never joined that protocol; its channel is the sitemap, which it re-reads on its own once the sitemap has been declared in Search Console.

A Jimp site now does both, in the same hourly pass that writes its unique text. When the schedule releases a video, the worker sends its address, and the home's, to IndexNow. One call, four engines. The site serves its own key, so there is nothing for you to set up and no webmaster account to open. Only what the schedule actually released goes out, never the whole catalogue: a fresh domain that announces ten thousand pages in one call is telling the engines exactly what it should not.

For Google, the worker declares the site's sitemap once in Search Console, with a service account of ours that owns the jimp.site property. After that Google re-reads it by itself, and every video in it carries the day it went up. We tested it on the six demo sites on 21 September: declared in the afternoon, read the same day, 201 to 364 pages discovered on each.

Custom domains are not covered yet. A domain of yours is its own property, which needs its own verification, and that is the next step. Until then its sitemap sits at the usual address and you can declare it yourself in a minute.

A channel pays from the first post

Search takes months to start paying, even when everything is told on time. A channel people follow pays from the first post: whoever follows it sees every release, and the link goes straight to the page. The trouble is that a feed only works if it is fed every day, and nobody feeds a feed every day by hand.

So the same hourly pass now posts every release on the site's Telegram channel and Bluesky account, once you have connected them in the site's Activity tab. Your channel, to be clear: your name, your avatar, your subscribers. Jimp's bot is a tool you add as admin of your channel, and it posts nowhere else; nothing goes to a channel of Jimp's. Telegram gets a card with the scene blurred, on purpose: Jimp's bot posts for every site from one account, and one explicit image would expose it to reports. Bluesky gets the sharp card with the adult label set by the author, which hides it from anyone who did not opt in. There is a daily cap you set, ten by default, one to forty-eight, because a feed that posts fifty times a day reads like a bot. And there is no catch-up: a channel you just connected gets the latest release and then the feed, never the whole back catalogue.

X and Reddit are not in there. X hides adult accounts from its search and charges for its API; Reddit throttles adult posting by API. We automate what works and say so; the rest stays manual.

The link itself has to look like something

There is one more place a new site loses clicks without noticing: the preview a link gets when it is pasted in a chat, a forum or a feed. Every network reads the same tags to draw it, and a generator that gives every site the same black card, or the raw thumbnail of a tube, throws that click away.

Each Jimp template now draws its own cards, from the site itself: its colours, its logo or wordmark, its fonts, its language, real images from its catalogue. One card for the home, one for every video with its rewritten title, performers and duration, and a soft twin of the video card with the scene blurred behind a play button. The soft one is what Telegram receives; the sharp one is what the networks read when someone pastes the link. There is no card editor, on purpose. The card is your site's, and it follows your site.

What this does not do

It does not rank your site. Nothing does, and anyone who sells that is selling weather. What it does is remove everything on your side that would have kept the engines from knowing: the weeks of silence, the pages never announced, the empty previews. The engines are told within the hour, every time, and whether they index a page stays their call, made on your niche, your domain and time.

It also runs only for subscribed sites, in the same pass as the unique rewrite. A demo shows you the site; the machine starts when you do. The whole of it, with every screen above and a few more, is on the page that tells how a Jimp site gets found.