What each figure counts
What every number on this site counts, what a missing one means, and what it cannot tell you.
Every number on this site is an observation of a public leaderboard in Last War: Survival, and each one counts something narrower than its name suggests. This page says what each figure counts, what it means when one is missing, and what it cannot tell you.
Three readings do most of the work. Same faces is a pair — how many of a warzone's fifty strongest commanders are still ranked on it, and how many still hold a top-fifty seat — and the pair inverts, because a board nobody climbs is a board whose seats do not turn over. Growth is the sum of positive gains, so its rank is the reading and its total is only a scale. Typical is the middle value, and it is printed beside the count it was taken over.
One habit runs through all six. A figure that has never been observed is shown as — rather than as zero, so a dash says something about the reading and nothing about the commander.
The words on every board
Warzone
The game's word for the numbered world a commander plays on. Every board and every map on this site is one warzone's.
Commander
The game's word for a player. One game account is one commander.
Alliance
A group of up to 100 commanders on one warzone, with a tag of up to three letters and a leader.
Tag
The short code in square brackets before an alliance's name, as the game prints it.
Roster
The members the game lists for an alliance, as it shows them. A roster is what has been seen, and it can be shorter than the member count the alliance reports.
Member
A commander on an alliance's roster.
Seat
One place on a board, or one place on a roster. "97 of 100 seats" is the roster; "#14" is a seat on a board.
Board
A game leaderboard for one warzone, ordered by one figure: power, kills, hero power, unit power or HQ for commanders; power, kills or members for alliances.
Ranked
Observed on a warzone's power board. A ranked count is the board's head.
Hero power
The power of a commander's heroes, as the game's hero board carries it.
Unit power
The power of the troops a base holds, as the game reports it. 0 is a real reading: a base holding no troops.
Building power
The power of a base's buildings, as the game reports it.
Kills
The game's kills figure for a commander, or the summed kills of an alliance's members. 0 is a real reading.
HQ
The level of a commander's headquarters, the game's own measure of how far a base is built.
Top power
The strongest commanders on a warzone's power board added together, at the depth the league states. A different quantity from any one commander's power, and the figure warzones are ranked by.
Filled
How many of an alliance's seats are taken, against the maximum the game allows it.
Tenure
How many days a member has been on a roster, counted from the joining day the game records for them.
Flag
The country code the game shows beside a commander. A flag is shown for some commanders and not others, and every share of flags is over the commanders that have one.
Active
Commanders whose power rose over the named window. Always printed beside the assessable count.
Assessable
The ranked commanders a warzone's change could be measured over: those observed on its power board at both ends of the window.
Marks
Everything the map shows on a warzone's ground: bases, and the warzone's own cities and alliance structures. Most marks are bases.
Window
The period a change is about: 7 days for activity and growth, 28 days for retention and for a commander's or alliance's change. A window is a length, stated as those days or as the days a record covers.
Record
The days a subject's history covers on this site. Change over a record shorter than 28 days is stated over the record's own length.
Season
The game's numbered era for a warzone, with a start, a settle and an end.
Cohort
Every warzone whose season started on the same day as this one's. Cohorts hold from a few dozen to a few hundred warzones and are the unit a warzone is compared inside.
Group
The eight warzones the game starts a season together, inside one cohort. A group is the game's rule; a cohort is a schedule.
League
Every warzone in one table, ordered by the power of its strongest commanders, at the depth the table states.
Trajectory
A warzone's seat in its cohort by power added over 7 days, and how many of the cohort added any.
Observed
A figure that has been seen on a board, a roster or a map. A figure never observed is drawn as a dash.
Power, and the figures beside it
What it counts. Power, Hero Power, Unit Power, Building Power, Kills, HQ and Members are the values the game's own boards carry for a commander or an alliance, kept as they were given. Ranked counts the commanders seen on a warzone's power board. Active counts the ones whose power rose over the past week, and the two are always printed together so the share can be recomputed from them.
Where a figure is a sum, the depth is part of its name. Top 100 Power is the hundred strongest commanders on a warzone added together, which is a different quantity from any one commander's Power, and the two are never drawn under the same bare word.
What a missing one means. A board figure of zero is shown as a dash, because a zero there is the absence of a row and never a measurement of the commander. Kills and Unit Power are the two exceptions: a commander who has killed nobody is a real zero, and so is a base holding no troops.
What it cannot tell you. Ranked is a count of the commanders on the boards, and a mark on the map is a commander's base, or one of the warzone's own cities and alliance structures; the two are different counts and a warzone's own page draws both. Read on 13 September 2026 over all 2,240 warzones in a published season cohort, how much of the top fifty's power the strongest alliance holds does not track how many of those fifty are still ranked four weeks later (ρ = +0.10, at a depth of fifty commanders), and a deeper read that makes the two look more related is moving with the number of alliance rows the board returned rather than with the warzones, which is why a share here always states the depth it was taken at.
Every commander on the boards is ranked by these figures on the commanders board, and how many each warzone has on its boards, beside how many bases are on its map, is on how busy each warzone still is.
Growth is the sum of positive gains
What it counts. Every positive change inside the window, added up. A negative change is dropped rather than subtracted, and a commander or a warzone with no baseline observation inside the window is left out rather than credited with its whole total. The windows are 24 hours and 7 days.
What a missing one means. A warzone with no gainers inside the window carries no growth figure at all, rather than a zero. A warzone that added nothing carries no rank either, because ranking it would place a dash seventh of nine.
What it cannot tell you. It is not net change. Power in this game does not fall, so a sum of positive changes over a live board is always positive: 64 of 64 gaining was true of every cohort on this site, and said nothing. The rank is the reading and the total is a scale. And the total is size-comparable only inside a cohort, whose members all started on the same day — a jump of the same size means different things on two warzones three seasons apart.
Where each warzone stands against the ones the game started it with is on how each warzone is moving.
Typical is the middle value
What it counts. The middle of the set: the value half the set sits above and half below. Every typical on this site is that. It is never an average.
It carries its own denominator. A middle is printed beside the count it was taken over, always, because a middle over 2,304 warzones and a middle over two are not the same evidence. Both terms of every ratio here are printed for the same reason — a reader who wants to recompute one has what they need.
What a missing one means. A middle of nothing is not zero. An empty set has no middle value, and the slot is a dash.
What it cannot tell you. It says nothing about the spread. Two sets with the same middle can be nothing alike, which is why a middle here is drawn beside the range it sits in wherever there is room for one.
The middles, and the spread each one sits in, are drawn for every warzone on how busy each warzone still is.
The group of eight and the season cohort
What each counts. A season group is eight warzones. The eight are the set a warzone meets in war mode. A season cohort is every warzone that started this season on the same day, which is what makes their boards comparable; a cohort holds 64 to 352 warzones, and the groups of eight are whole slices of it.
What a missing one means. Where the game does not state a warzone's season fully, it carries no group and no cohort rather than a guess at either, and the warzone is simply absent from the cohort's table.
What it cannot tell you. It is not a statement about who may transfer where. The game does not publish that rule, and this site has never established it. Starting together is what makes two boards fair to compare; it is not a claim about where a transfer ticket will take you.
Every cohort, with its groups of eight inside it, is on the season matchups.
Same faces is a pair
What it counts. Of the commanders a warzone held at ranks 1 to 50 four weeks ago, how many are ranked on it now, and how many still hold a top-fifty seat. Two numbers, drawn together, never one on its own. The window is 28 days, anchored to the last complete day rather than to the day in progress — a figure built on a day still filling could only rise until midnight and then reset.
Which quadrant a warzone lands in is the reading. A commander who was out-grown is still on the board and counts in the first number but not the second; a commander who left counts in neither. So a low first number beside a high second is a warzone people left, and a high first beside a low second is a warzone people are still climbing.
What a missing one means. The pair needs two witnesses: a complete fifty-row board four weeks ago, and a board that answers in full now. A warzone missing either carries no reading at all rather than a zero, because 0 of 50 · same faces on a page somebody is choosing a warzone from is the worst possible thing to print about a board that did not answer.
What it cannot tell you. The seat count alone is not a health reading. A board nobody climbs is a board whose seats do not turn over, which is why it is never drawn without the count beside it. And growth does not stand in for it: across all 2,240 warzones in a published cohort, power added over the week and seats held move together at ρ = −0.001, which is to say not at all, with the strongest and weakest cohorts running from −0.635 to +0.396.
Every cohort's boards, each with its own pair, are on the season matchups.
Shield states, and the power on a map row
What a shield reading counts. Open is the game drawing no marker at all — the state is the missing bubble, and open is the word players use for it. Shielded carries the game's own stated expiry. Lapsed is the third, and it is a claim about the reading rather than about the base: a base seen with a standing shield whose stated expiry has since gone by. Nobody watched that bubble come down, the game may have replaced it the moment it fell, and on an archived pass that gap is the whole age of the pass — which is why it is drawn dimmed.
What the power on a base row counts. A dotted figure is the commander's own power, from their alliance's member roster — the same figure their dossier carries.
What a missing one means. Shield — not reported is an absent reading, and it is never drawn as open, which is a different claim. The power column has three absences and three sentences, because they mean three different things: power — not reported, power — not looked up for this warzone, and power — looking it up.
What it cannot tell you. Whether a base is shielded right now. Every time drawn here is the game's own, and a shield can be replaced at any moment. A row with no power figure carries no claim about that commander's strength, and a power filter cannot include those rows at all — the miss clusters by alliance, so a filtered list is not a ranking of the warzone.
Both readings are drawn on every base of a warzone's own field on the map workspace.
One reading, in full
Read on 13 September 2026.
All 2,240 warzones in a published season cohort carried the same-faces pair; none refused. The middle board kept all fifty of its strongest commanders ranked and 33 of the fifty seats. 1,229 boards kept all fifty ranked and 17 kept ten or fewer. Split at those two middles, 891 boards are high on both and 665 are low on both, while 346 hold a high seat count on a low ranked count — boards people left — and 338 fall the other way, which is what a board people are still climbing looks like. Still ranked and seats held move together at ρ = 0.537; power added over the week and seats held, at ρ = −0.001.
One cohort, in the same reading. On the 128 warzones that started season 6 together on 29 June — numbers 805 to 932 — all 128 answered. The middle board kept 40 of its 50 strongest commanders ranked and 24 of them in a top-fifty seat. Twelve boards kept a high seat count on a low ranked count: people left those. Eleven kept a high ranked count on a low seat count: people are still climbing those. And of the ten boards holding the fewest seats, four sit in the top half of the same 128 by power added over the week — a board can add power because its survivors are grinding, or because arrivals are filling the seats the old fifty vacated, and growth alone cannot tell those apart. That cohort's own page is warzones 805 to 932, where the figures have moved on since; the numbers above are the reading taken on the day this page names.
Three counts sit close together in that paragraph and they are not the same count: 2,321 warzones in the catalog, 2,304 we can assess, and 2,240 in a published season cohort on that day. The 81 outside the cohort reading are the newest contiguous block, numbers 2,245 to 2,325, which no published cohort holds yet.
The record starts on 7 July 2026. Nothing here reaches further back than that, for any warzone, commander or alliance.