What the Meter Score is
The Meter Score is a 0–100 rating computed for all 1,329 sites in the directory. It is a weighted sum of nine metrics, each measured on its own 0–100 scale. The formula is identical for every site: no category gets a custom formula bolted on, and nobody at PornMeter can manually move a site up or down. When a score changes, it is because the underlying measurements changed in that week's crawl.
The weights favor the qualities that make a site worth returning to — catalog depth, video quality and trust — over the qualities that make a site merely tolerable. A site with a spotless popup record but a shallow library will not top a major category, and a huge catalog wrapped in aggressive ads will not either.
The nine metrics and their exact weights
| Metric | Weight | What it measures |
|---|---|---|
| Library depth | 18% | How much content the site actually has: catalog size, category coverage and how deep the long tail goes. |
| Video quality | 16% | Best available resolution and how consistent quality is across the catalog, not just on the front page. |
| Freshness | 12% | How often new content lands and whether the site is still actively maintained. |
| Trust & safety | 12% | Valid HTTPS, sane redirect behavior, absence of malware-pattern ad networks, working legal pages and content-removal process. |
| Player & UX | 10% | Does the player start fast, seek reliably and stay free of fake buttons and forced mid-roll interruptions? |
| Mobile experience | 10% | The same site loaded on a phone-sized viewport: layout, tap targets, player behavior and mobile ad pressure. |
| Low ad pressure | 8% | How much advertising interferes with actually using the site. |
| Popularity | 8% | Real-world usage as a sanity signal, so obscure sites with clean tests do not outrank sites millions of people successfully use daily. |
| Free access | 6% | The share of the catalog you can watch without paying, shown as the ~% free chip on reviews. |
Library depth (18%). How much content the site actually has: catalog size, category coverage and how deep the long tail goes. Estimated from crawled listing pages, sitemap coverage and category counts. Bigger is not automatically better — the score also reflects whether the catalog is navigable.
Video quality (16%). Best available resolution and how consistent quality is across the catalog, not just on the front page. We sample player manifests across sections. 4K and VR claims are only marked verified when we can confirm them in a real playback session.
Freshness (12%). How often new content lands and whether the site is still actively maintained. Measured from upload timestamps and week-over-week diffs between crawls. When a site hides dates, freshness is estimated from catalog churn and flagged with lower confidence.
Trust & safety (12%). Valid HTTPS, sane redirect behavior, absence of malware-pattern ad networks, working legal pages and content-removal process. Checked on every crawl. A single failed check never zeroes the score, but repeated failures sink it quickly.
Player & UX (10%). Does the player start fast, seek reliably and stay free of fake buttons and forced mid-roll interruptions? Measured in a scripted real-browser session: time to first frame, seek latency, overlay count on the player surface.
Mobile experience (10%). The same site loaded on a phone-sized viewport: layout, tap targets, player behavior and mobile ad pressure. A separate mobile crawl produces the Mobile score you see on review pages (0–100). Sites that redirect mobile users to a different, worse domain are penalized.
Low ad pressure (8%). How much advertising interferes with actually using the site. We count popups, popunders, overlays and third-party ad requests during a scripted visit, then grade ad load on a 0–3 scale. The grade is inverted for scoring: zero interruptive ads scores 100, each level of aggression removes 25 points.
Popularity (8%). Real-world usage as a sanity signal, so obscure sites with clean tests do not outrank sites millions of people successfully use daily. Blended from public traffic estimates. Deliberately the smallest quality weight — popularity confirms, it never drives.
Free access (6%). The share of the catalog you can watch without paying, shown as the ~% free chip on reviews. Estimated by sampling content across sections and checking what actually plays without an account or payment.
How the weekly crawl works
Every site is re-crawled on a weekly cycle. The crawl has two parts. The first is a structural pass: listing pages, category trees, sitemaps and upload dates, which feeds library depth and freshness. The second is a scripted browser session — a real Chromium instance, not an HTTP scraper — that loads the homepage, opens a video page and interacts with the player the way a person would.
During the browser session we record the numbers you see quoted on review pages: popup and popunder count, overlay count on the player, number of third-party ad requests, largest contentful paint (LCP), and time until the video actually starts playing. The same session is repeated in a phone-sized viewport to produce the separate mobile score. The date of the last completed test is shown as the "Tested" chip on each review.
Observed vs estimated data
Not everything can be measured directly on every site. When a site hides upload dates, freshness is estimated from how much the catalog changed between crawls. When a paywall blocks catalog sampling, free share is estimated from the visible tier. Estimated values carry a lower internal confidence weight, and where a review relies on an estimated number, the text says so in plain language rather than presenting it as a measurement.
Feature claims — 4K, VR, downloads — use a three-state model: verified (we confirmed it in a session), absent (we confirmed it is not there), or unverified (we could not check). Review pages only assert features from the first two states; unverified features are either omitted or explicitly marked as unconfirmed. Review text is generated from this fact sheet and is not allowed to add features from memory.
Missing data
If a metric cannot be measured or estimated at all, the site is scored on the metrics we do have, re-normalized so it is not silently punished for a gap in our pipeline. Sites that fail the crawl entirely (offline, geo-blocked for our probes, or blocking automation) keep their last known score for a grace period and are then dropped from rankings until they can be measured again.
How category rankings work
A site's position inside a category (the "#3 of 45" chip) does not simply reuse the global Meter Score. Each category re-scores its members on the factors that matter for that intent: the free-tubes ranking leans on free access, ad pressure and player behavior; live cams on free-chat generosity, room load and pricing transparency; premium networks on trial terms, exclusivity and 4K delivery; AI tools on output quality, speed and privacy. The factor table on every category page ("What we tested for this ranking") lists exactly what was weighed for that list.
Update cycle
Scores recompute after every weekly crawl; the current data week is shown at the top of this page and in the site header. Rankings can and do move week to week — that is the system working, not noise. Larger shifts usually trace to a site redesign, an ad-network change or a measured outage.
Affiliate links and independence
PornMeter earns money from affiliate links on some outbound "Visit" buttons. All outbound site links go through our /go/ redirect and carry rel="sponsored nofollow". The affiliate system and the scoring pipeline are physically separate: the crawler does not know which sites have affiliate deals, and the score formula has no input for revenue. A site with no affiliate program ranks exactly as its measurements dictate — several category leaders on this directory pay us nothing.