Public-data scoring · BFR

How the BFR score works

Six public-data factors help readers compare and shortlist creators. The score does not rate the private subscription experience.

Documentation updated 20 September 2026.

What we measure

The current directory model uses public listing snapshots collected from OnlyFinder: display name and bio, listed subscription price, photo and video counts, likes, and directory category membership. Software calculates the factors within each niche. We do not subscribe to every account, test private messages or verify paid-content quality.

Community stars are separate reader ratings. Sponsored featured placements are advertising and do not determine the editorial score.

Six factors, with explicit limits

Maximum factor points sum to 100. The total is divided by 10 for the displayed score. The current weights are 20 / 15 / 15 / 20 / 15 / 15. Earlier text listed different weights and overstated what we tested; this documentation corrects those claims.

Profile clarity · 20 points

A name-and-bio heuristic combines a base value, a name-specificity rule and bio length, capped at 220 characters. It does not establish that a name is real, verify identity or assess the accuracy of a biography.

Content library · 15 points

The public photo-plus-video count is compared with niche peers and a logarithmic volume scale. Some scorecards call this “Posting activity”. The input measures library size, not posting frequency, recent activity or gaps between posts.

Price & value · 15 points

Free subscriptions receive 13.2 points. Paid subscriptions are scored relative to listed paid prices in the niche, with lower prices receiving more points. This does not measure paid extras, tips, private-message costs or subscription quality. Check current prices before subscribing.

Niche fit · 20 points

The model favours profiles appearing in fewer directory niches, checks for the niche phrase in the display name or bio, and applies an age-consistency adjustment in numeric-age categories when age information is available. Directory categories and profile claims can be wrong. These are matching signals, not a verified content review or proof of age.

Engagement · 15 points

Total public likes divided by combined photo and video count produces a ratio, compared with niche peers and a logarithmic scale. Legacy data calls this “Interaction quality”. It does not measure replies, DM quality, subscriber satisfaction or renewal rates.

Reach · 15 points

Public likes are compared with niche peers and a logarithmic scale. Legacy data calls this factor “Trust signals”. Likes are not unique subscribers or followers; this factor does not verify identity, safety or trustworthiness.

Calculation

BFR score = (clarity + library + price + niche fit + engagement + reach) / 10
Maximum points: 20 + 15 + 15 + 20 + 15 + 15 = 100

Individual factors are rounded to one decimal before summing. A creator can have different scores in different niches because the comparison group changes.

What are the numeric formulas?

Let P(x) be the fraction of eligible niche profiles at or below x; L(x) = log10(max(1,x)); C(x) clamps to 0–1. Reach = 15 × [0.6P(likes) + 0.4C((L(likes)−3)/4)]. Engagement = 15 × [0.65P(likes/items) + 0.35C((L(likes/items)−0.7)/2)], using max(1,items) as denominator. Library = 15 × [0.5P(items) + 0.5C((L(items)−1)/3)]. Paid price = 15 × C(0.45 + 0.45 × (1−P(price))), with paid profiles as peers. Niche fit = 20 × C(0.45 + 0.40 × C(1−L(niche count)/L(250)) + keyword adjustment + age adjustment). A literal niche-phrase match adds 0.12. In numeric-age niches with age data, the adjustment is 0.15 within two years, 0.05 within five years and −0.1 otherwise. Clarity = 20 × C(0.25 + 0.40 × name heuristic + 0.35 × C(bio characters/220)). The name heuristic is 1 for the configured name-like capitalisation rule, 0.55 for other non-generic names of at least three letters or digits, and 0.25 otherwise. These language-sensitive rules have limitations.

Category summary figures

NSI · niche-fit summary

Mean niche-fit points in the ranked set, divided by 20 and multiplied by 100. This summarises the matching heuristic, not verified niche authenticity.

EQS · average editorial score

Mean displayed BFR score in the ranked set, out of 10. When starting from internal scores out of 100, divide the mean by 10 once.

FTC · free-tier coverage

The percentage of the ranked set listed as free to subscribe in that snapshot. Paid extras may apply.

RMI · legacy engagement band

Some older pages express engagement bands as “Minutes”, “Within 1 hour” or “Within 24 hours”. These are not measured reply times and should not predict a response. BFR has no private-message timing data. Thresholds use mean engagement points: at least 10, 7 or 4 respectively; below that, “Variable”.

Eligibility and corrections

The model applies matching filters, including age-claim and maturity-keyword rules in relevant niches. They do not independently verify age or detect every unsuitable profile. The earlier promise of automatic detection of banned accounts, impersonation, 60-day posting gaps or a low-factor exclusion was unsupported. Report wrong or inappropriate listings through Contact; do not send private or explicit material.

Dates and updates

Scores describe a dataset snapshot, not a live subscription audit. Updating documentation does not mean every creator was rechecked. Review dates must follow actual work; they must not advance automatically each Monday. We do not promise weekly re-audits of every top creator or a 48-hour refresh.

Reproduction and limitations

Reproduction requires the same input snapshot, eligible niche population and scoring version. A current public profile alone is insufficient because prices, counts and comparison groups change. For disputed results, send the page URL, handle and public source to contact@bestonlyfansreviews.com. See Sources.

Questions

Do you test paid content or private messages?

No. Directory scores use public listing snapshots. They are not first-hand subscription reviews or measured response-time claims.

Can payment improve an editorial score?

No. Sponsored featured placements are separate from the six-factor score and editorial ranking.

Are community stars part of the BFR score?

No. Reader ratings are displayed separately from the public-data editorial score.

Correction record

20 September 2026: Corrected factor weights and unsupported testing, identity-check and refresh claims. This changes documentation, not stored creator scores.

Niche-fit-first category pilot

From 28 September 2026, the English Goth Girl, Redhead and Fitness selections use niche fit at 50% and the five remaining editorial factors at 10% each. The twelve picks in each selection have a niche-related reference in their saved profile descriptions. This selection step is separate from the existing factor calculation; it is not an exhaustive review of every directory record or verification of paid content. Source snapshots date from 4 June 2026. Other categories retain their existing method until reviewed.

Current-profile pilot · 28 September 2026

Goth Girl, Redhead, Fitness, Cosplay and Tattooed each use twelve current public profiles. Bios and public price/photo/video/like counters are retrieved together. A relevant self-description is required for eligibility; old directory tags alone are insufficient. Renamed/unavailable profiles, missing metrics, unrelated promotional mentions and announced hiatuses are excluded.

Niche fit is a binary eligibility factor worth 50 points. Profile clarity, library size, price & value, likes per media item and total likes are worth 10 points each. Likes use 60% shortlist percentile and 40% log-scaled magnitude (1,000–10 million). Likes per media item use 65% percentile and 35% log magnitude (approximately 5–500). Library size uses equal percentile and log magnitude (10–10,000). Free subscriptions receive 8.8/10 for price; paid subscriptions receive 4.5–9 according to their price percentile among paid picks. Clarity gives 5 points for a retrievable public profile and up to 5 for bio depth (220 characters). Values are capped to each factor range and rounded to one decimal before summing.

The resulting score describes public data within the checked shortlist; it does not verify paid content quality, reply times, subscriber counts or posting frequency. Self-descriptions are attributed to their authors. Every displayed bio is complete, including disclosures. Other directory totals and unrefreshed related listings keep their historical dates. This content intervention must be excluded or separated when analyzing the title experiment.