How the ranking works
viralsaas.lol is a public leaderboard of SaaS landing pages. Three AI judges score every submitted page against the same 100-aspect rubric, and the page then earns its position by being compared head to head against sixteen hidden reference sites. The $5 buys the evaluation. It does not buy the position, and there is no price at which it would.
The order comes from duels, not from scores
Every submission is put up against a set of hidden anchor sites, one pair at a time. Each comparison is run in both directions so that being shown first cannot help a page win. The wins and losses are fitted with the Bradley-Terry model — the same method used to rate chess players from game results — which turns them into a single strength value. The board is sorted by that strength.
The number shown next to each listing is that strength on a fixed 0–100 scale. It cannot disagree with the ordering, because it is the quantity the ordering is sorted by. It is not a percentage and not a grade: it is a position on a ladder whose rungs are real distances, which is why the gap between #1 and #2 is small and the gap between #2 and #3 is not.
The rubric grades the page. It does not decide the rank.
The hundred aspects cover clarity, pricing, trust, content, design, mobile, speed, technical hygiene, onboarding, differentiation and accessibility. Thirty-two of them are measured by script rather than judged — load time, security headers, alt-text coverage, console errors — and handed to the judges as facts they are not allowed to contradict. The rest are scored by three independent AI vision models, each of which sees the same screenshots and the same page text.
That produces a rubric score, published on every listing page beside the categories that justify it. It is genuinely useful and it is genuinely not the ranking. A page can tick a great many boxes and still lose when a judge is asked which of two pages is better — which is exactly what the duels ask. On this board the site at #2 has a higher rubric score than the site at #1. That is not a bug being tolerated; it is the reason the ranking cannot be gamed by working through a checklist.
Why the reference sites are hidden
Sixteen anchor sites form the fixed reference every submission is measured against. They are never listed and never appear on the board. Keeping them hidden is what stops a page being tuned against the specific opponents it will face, and holding them constant is what makes a position earned today comparable to one earned months ago.
What the $5 does and does not buy
It buys one evaluation of one landing page: three judges, the full rubric, the captured evidence, a written verdict per category, a share card, and a permanent public listing. It does not buy a position, a boost, a review, or removal from the board. Everyone pays the same amount and is judged against the same reference set, and the sites at the bottom of the board paid exactly what the sites at the top paid.
An earlier version of this product sold rank to the highest bidder. It was retired, and the rule it broke is now the one thing this board exists to protect.
Questions
Can I pay for a better position on viralsaas.lol?
No. The $5 buys one evaluation: three AI judges score the page and it is duelled against the anchor set. The position that comes out is whatever the duels produce. There is no paid placement, no boost, and no way to buy a higher rank at any price. Every listing on the board paid the same $5, including the ones at the bottom.
How is the ranking order decided?
By head-to-head duels. Each site is compared against a set of 16 hidden anchor sites, both ways round to cancel position bias, and the wins and losses are fitted with the Bradley-Terry model to produce a strength. The board is sorted by that strength and by nothing else.
Why does a site with a higher rubric score rank below one with a lower score?
Because the rubric does not decide the order. The rubric grades a page against a fixed checklist: is the pricing clear, is there trust evidence, does it load quickly, does it work on a phone. The rank comes from a different question: shown these two pages side by side, which one is better? A page can satisfy a lot of checklist items and still lose that comparison. This happens on the live board — the site at #2 scores higher on the rubric than the site at #1.
What are the anchor sites?
Sixteen well-known SaaS landing pages that are never listed publicly and never appear on the board. They are the fixed reference every submission is measured against, which is what makes positions comparable between sites judged weeks apart. They are hidden so that nobody can optimise a page against the specific set it will be compared to.
Who are the three judges?
Three different AI vision models from three different vendors, chosen so that no single model's taste decides an outcome. Each one sees the same captured evidence — full-page screenshots at desktop and phone width, the page text, the pricing, and a set of scripted measurements — and scores the same 100 aspects independently. Their individual scores are published on every listing page.
How long does an evaluation take, and when does my rank appear?
Judging usually finishes within a few hours of payment, longer when the queue is busy. Positions become public at the daily reveal, 00:00 UTC, so a site judged during the day appears on the board that night. Progress is visible on your dashboard the whole time.
What is measured by a script rather than judged by an AI?
Thirty-two of the hundred aspects are measured, not opinions: HTTPS and security headers, load timing, page weight, console errors, alt-text coverage, heading order, form labels, viewport and mobile overflow, favicon and Open Graph tags, link health, and the 404 page. Those measurements are shown on every listing page and the judges are given them as facts they cannot contradict.