# How similar channels are selected

“Similar channels” are the channels it is fairest to compare a given channel with: the same topic, the same region, a similar size, shared links and reposts. They power the “Similar channels” tab, the “among similar” comparisons on other tabs and the similar channels in the channel card. This page explains the selection rules and how comparison metrics are calculated.

## Where you see it

- The **“Similar channels”** tab of a channel page — the table of similar channels, “Channel vs the median of similar channels”, rank among similar and other blocks. Details — [Similar channels](/en/channel/similar/).
- “Among similar” blocks on other tabs: the channel's rank by reach, engagement, ad share, the dashed line of similar channels on charts. For example, [Post reach](/en/channel/reach/), [Ad statistics](/en/channel/ads/).

## Who we consider

Candidates are only **public active channels** (not chats) with **100 or more subscribers**. There are four sources:

1. **Same topic** — the same primary category and a size no more than 5 times different (up to the 150 largest).
2. **Same region** — a channel from the same region or its cities (only if the channel's territory is a region or a city, not a country), size no more than 5 times different (up to the 100 largest).
3. **The channel mentions them** — channels it linked to or reposted during the last year (up to the 200 most frequent).
4. **They mention the channel** — channels that linked to it or reposted it during the year (up to 200).

## How similarity is scored

Each candidate gets **similarity points** — at most 100:

| Signal | Points |
|---|---|
| Same primary category | 35 |
| Otherwise — at least one shared category | 20 |
| The channel mentions the candidate (n times) | 10 × ln(1 + n), at most 25 |
| The candidate mentions the channel (n times) | 6 × ln(1 + n), at most 15, × size closeness |
| Mentions in both directions | +10 |
| Size closeness | 15 × closeness |
| Same region | 10 |

**Size closeness** — how similar the channels are in subscriber count: identical channels — 1, a 10× difference — 0.5, 100× or more — 0. So links from small channels to a big one weigh almost nothing: a school channel with 700 subscribers that linked twenty times to a national news outlet with millions of subscribers is not similar to it.

Channels with **5 points or more** enter the selection, at most the **50** most similar. Under each similar channel's name it says why it was picked: “same topic”, “same region”, “mentions this channel”, “this channel mentions it”, “mutual mentions”, “similar size”.

!!! note "If the channel has no category"
    Topics are assigned by moderators and owners (details — [Categories and territories](/en/metrics/categories/)), and not every channel has one yet. For a channel without a category, similar channels are found by mentions and region. Verify your channel and set its category — the selection becomes more accurate.

The selection does not depend on the chosen period and is refreshed every 6 hours.

## Comparison metrics

The comparison table and tiles are calculated for the tab's selected period — full days up to and including yesterday (today's posts are still collecting views):

| Metric | How it is calculated |
|---|---|
| Subscribers | How many now |
| Growth | Subscribers now vs the first day of the period, in percent |
| Views per post | Views of posts published in the period ÷ their number |
| ER | Reactions ÷ views of the same posts × 100 % |
| CI | [Citation index](/en/metrics/citation-index/) — always for the last 30 days |
| QI | [Quality index](/en/metrics/quality-index/); if not calculated — “—” |
| Posts per week | Posts in the period ÷ days in the period × 7 |

**The median of similar channels** is the value that splits similar channels in half: half have more, half have less. In the “Channel vs the median of similar channels” block each tile shows the channel's difference from the median: for growth — in percentage points, for the rest — in percent of the median. The strip at the top of a tile is green if the channel is not below the median, otherwise red.

**Where the channel is strong and where it lags** — the channel's rank on each metric among similar channels: green — the top third, yellow — the middle, red — the bottom third.

**Neighbours in the ranking** — the channel's rank by subscribers in its city, region, federal district, topic and all of MAX: three above and three below. Ranks are recalculated every 6 hours.

## Who sees what

The similar channels table (first 11 rows), “Channel vs the median of similar channels”, the first pairs of “Most frequent links” and “Neighbours in the ranking” are open to **everyone**. Sorting the table by any metric, filters by selection reason, the traffic light and some blocks — **after signing in (free)**. “Rank among similar” and “When similar channels post” — **for the channel owner and with a subscription**. All similar channels instead of 11 rows — **with a “Premium” or “Business” subscription**. Details — [Similar channels](/en/channel/similar/).

## See also

- [Similar channels](/en/channel/similar/) — what is on the tab and how to use it.
- [Categories and territories](/en/metrics/categories/).
- [Quality index](/en/metrics/quality-index/) — it also compares a channel with others of the same topic and size.
- [How metrics are calculated](/en/metrics/basics/).