Transparency

Ranking methodology

The RankUp rating is calculated only from members' published reviews and star ratings. The formula, its parameters and the version history are published here. Paid ads are not part of the calculation.

Current version: community-v2

In short

A brand's rating is a Bayesian average of members' 1–5 star ratings.
A review by an author with a verified e-mail counts in full; any other review counts with weight 0.25.
As a review ages, its weight halves every 365 days, but never drops below 0.25.
A brand gets a rating once it has 3 reviews; brands with fewer are shown as a separate group.
Likes don't affect the rating — they only count in the “Popular” order.
Ad placements are shown separately and don't change the community ranking.

Formula

A brand's rating R is calculated as:

R = (C·m + Σ wᵢ·rᵢ) / (C + Σ wᵢ)

rᵢ — the star rating in a review, from 1 to 5.
wᵢ — the review's weight (below).
m — the average rating: the category's own once it has at least 30 reviews, otherwise that of all categories. With no reviews at all, m = 3.0.
C = 5: every brand starts as if it had 5 reviews rated m, so one or two reviews can't swing its rating.

Review weight

wᵢ = verification × recency

Verification: an author with a verified e-mail — 1, other authors — 0.25. Telegram or SMS verification will be added in a later phase.

Recency: 0.5 to the power of (the review's age in days / 365), but at least 0.25.

Example weights
Review age	Verified author	Unverified author
New	1.00	0.25
6 months	0.71	0.18
1 year	0.50	0.13
2 years or older	0.25	0.06

When a brand gets a rating

A rating needs at least 3 qualifying reviews: a verified author's review counts as 1 and any other as 0.25 (for example 3 verified or 12 unverified reviews). A review's age doesn't count here, so a brand never loses its rating just because time passes. Brands that don't have enough reviews yet are shown after the rated brands, in the “Not enough reviews yet” group.

Order

Rated brands first, by rating (to two decimals).
On equal ratings, the brand with more weighted reviews.
If that is equal too, the brand with the more recent latest review, then by name.
Then the brands without a rating yet, by popularity, then by name.

Pages show the rating rounded to one decimal; the order uses the value to two decimals.

The “Popular” order

Popularity = counted likes + weighted reviews. It is a separate order and is not part of the rating. Each member can like a brand once.

Worked example

The category average is m = 4.0. A brand has 3 new 5-star reviews from verified authors: R = (5 × 4.0 + 3 × 5) / (5 + 3) = 4.38.

What doesn't count

Paid ad placements and auction bids: the ranking module doesn't read them at all.
Reviews hidden by a moderator.
Removed likes, and suspicious likes held for review because they came too fast.
Votes from company and admin accounts: they can't like or review brands.
Manual changes: nobody, including the RankUp team, can change a rating by hand.

Version history

Every day a copy of each category's ranking is stored with its version label and never changed afterwards. Positions from different versions are never compared with each other.

Version	In use	What it does
community-v2current	since Nov 5, 2026	RankUp rating: weighted Bayesian average, a rating from 3 reviews, “Popular” as a separate order.
community-v1	Nov 4, 2026 – Nov 5, 2026	Activity points: each like 1 point, each review 2 points. Star ratings were not used.

See the rankings Advertising policy