How we test
Every site gets the same tests, run by the same scripts, from a UK home-broadband connection, every week. Nothing is scored by hand except where noted, and every number links back to the raw evidence we keep.
The searches
- Home, England, London, Sedona, Arizona (a small US town of about 10,000 people), Asian and Red hair — using the site's own pages where they exist, otherwise its search.
- Combined filter search: Sydney + blonde + young + female + cosplay, using the site's real filter controls.
What we measure
- Live and fresh: every creator shown in the first 200 organic results is fetched by our own profile scraper. Live = the profile exists. Fresh = last seen, or new posts/photos/videos, within 14 days. The check date is stored for every profile.
- Paid placements: any listing that routes through a tracking or campaign link (for example
/c15-style OnlyFans campaign links or ad-network redirects) or is flagged as paid in the site's own data counts as sponsored. We then check whether the card is visibly labelled. We never click ad links to find out — usernames come from the ad data itself. - Filters: how many of the five combined-search criteria a visitor can apply through the site, and whether they actually change the results.
- Search speed: time from search to first result, three runs each, with our own connection time measured separately.
Location accuracy: a result only counts as right or wrong when the creator's location can be judged. A profile that only says "United Kingdom" can't be checked for London, and text such as "Traveler" or "Closer than you think" isn't a place. Both count as "location not disclosed", the same for every site.
Score weights
| Component | Weight |
|---|---|
| Fresh results | 20 |
| Page coverage | 10 |
| Live profiles | 10 |
| Ads labelled | 15 |
| Ad density | 10 |
| Filters that work | 15 |
| Search speed | 5 |
| Tag accuracy | 15 |
| UK access | 10 |
| Location accuracy | 15 |
| Page speed | 10 |
| Features | 15 |
While some tests are still pending, the score uses only the components already measured, re-weighted to 100. Weights are fixed and published before each run.