How to upgrade a precon Commander deck: what 55 decks say to change first

We measured 55 Commander precons. Most already meet the usual role counts, so here is how many cards to change, in what order, how to pick the cuts, and one real deck priced swap by swap.

Aymeric Lamontagne

Nov 6, 2026

The advice you will usually hear about upgrading a precon is to add ramp, add removal and add card draw. We counted what 55 recent Commander precons actually contain, and the honest finding is that the typical precon already meets those counts. The median list holds 38 lands, 13 ramp pieces, 14 draw cards, 10 removal spells and 3 board wipes, and 23 of the 55 clear all four role targets our Agent starts from, with no deck short of lands.

Where the lists lag is speed and fit. Nine of a typical multicolour deck’s 38 lands enter tapped or fetch a tapped land, a quarter of its spells cost five mana or more, and only three of its answers cost three or less at instant speed. And about 25 of its 61 spells do none of those four jobs, which is where the deck’s plan lives and where no counter can do your reading for you.

So the short answer: change about a dozen cards in a first pass, cheapest first, then play before changing more. For the deck worked through below, twelve swaps came to $9.33 on November 2, 2026.

This piece covers what the lists contain, how many cards to change and why, the order to change them in, how to choose the cuts, one real deck carried through all of it, and what small budgets buy.

Lands

38 — None of the 55 below 36

Slow lands

9 — Tapped, or fetch a tapped land

Spells at 5+ mana

15 — Of 61 nonland cards

Twelve swaps

$9.33 — One real deck

Medians across 55 Commander precons released February 9, 2024 to October 2, 2026. The slow-land median covers the 50 multicolour decks. The last tile is the Silverquill Influence package below, priced November 2, 2026.

The method in six steps

1. Write the plan in one sentence, from the commander’s own text, and play the deck as printed if you can.
2. Count, then read. Compare lands and roles against a starting target, then read the cards the count credited.
3. Replace slow lands that do nothing else. A basic land is untapped and nearly free.
4. Fill the gaps the reading found: cheap mana first, then cheap answers.
5. Replace what fails the plan test, and spend what is left on cards that serve the plan twice.
6. Stop and play. Expensive lands come last, if they come at all.

The rest of this piece is why it runs in that order.

What a precon actually contains

The sample is every regular Commander precon in our catalog released from February 9, 2024 to October 2, 2026, one row per list: 55 decks, leaving out Collector’s Editions and the Star Trek decks, which do not release until November 13. Role counts use the rules behind the quota tiles in the builder’s Stats panel, read on November 4, 2026. The targets are the starting quotas our Agent reads: 36 to 38 lands, 10 ramp, 10 draw, 9 removal, 3 wipes.

Add up each deck’s shortfall across the four spell roles and the median is one card. Twenty-three decks miss nothing, twenty-three miss one role, and nine miss two or more. A precon may have other problems, but for most of them a shortage of categories is not the main one.

Two caveats before anyone relaxes. First, the counter is generous: it reads card text and Scryfall Tagger’s labels, so it credits a conditional draw effect the same as a clean one, and the worked deck below shows it counting an eight-mana Aura as removal. Second, a target is a prompt for review, not a finding. A deck that averages under three mana does not need what a deck averaging four does.

Speed is where the lists look weaker:

The first two rows are what we will call slow lands. Across the 50 multicolour decks the median is nine, with the middle half between six and eleven; the five mono-coloured Foundations Commander decks hold one to three, because they have fewer colours to fix. With nine slow lands among 38, the chance that at least one of your first four land drops is slow works out to 68%, and two or more to 23%.

The measure we would most like is the one we cannot compute honestly. “Cards with no synergy” needs a plan, the plan lives in the commander’s text, and a mechanical check would be guesswork presented as data. What we can count is the room for it: about 25 of a median 61 spells do none of ramp, draw, removal or wipes. Those slots carry the plan, and the only reliable test of them is the one you run by hand, below.

How many cards to change when upgrading a precon

About a dozen in the first pass. That number is an estimate from two kinds of arithmetic, because nothing in this data records who won.

The amount of work. Taking slow lands from nine to four is five swaps. The median role gap is one card, and reading rather than counting usually finds one or two more. The plan test on our worked deck turned up five clear cuts and two thin ones. That comes to somewhere between ten and fourteen. Our Wretched Ranks guide reached twelve from a different direction, by asking what its commander’s sacrifice engine lacked; it is one more deck, not proof.

What a game can tell you. On the draw, you have seen 15 of your 99 cards by turn eight, before any card draw. So any one card turns up in about 15% of games, roughly one in seven, and a single swap is nearly impossible to judge from a few sessions. Twelve new cards means about two of them per game: enough to notice whether the deck feels different. Forty means about six per game, with no way to tell which of them did what.

If you like the deck as it is, change fewer. If you are replacing most of those 25 plan slots as well as the mana, you are building a different deck, which is a perfectly good thing to do, and worth calling a rebuild so you pick its commander on purpose.

The order of operations, and why

Write the plan, then count and read

Everything later depends on one sentence saying how this commander wins, because the cut test cannot run without it. Then compare the deck with the targets and read what got counted. The Stats panel gives you the count in seconds, and our guide to reading it explains each tile. If you have an account, test a few opening hands in Playtest before you buy anything: that is how you find out whether the deck is short on mana or only looks it.

Replace the slow lands that do nothing else

Lands come first for two reasons, and neither is that they are the biggest fix. They cost almost nothing, and a land swap takes no spell out of the deck.

The first candidates are the ones the catalog keeps reprinting. Among the 55 decks, Path of Ancestry appears in 33, Evolving Wilds in 29, Terramorphic Expanse in 27 and Temple of the False God in 17. That last one produces no mana at all until you control five lands. Each of the others enters tapped or fetches a tapped land, which costs nothing on a turn when you had no play and a turn’s mana on one when you did.

Here is the deflating part. We ran a small model of land drops and mana rocks, described in the note below. In a 38-land deck with three two-mana rocks, moving from nine slow lands to four lifts the chance of having five mana on turn five from 59% to 66%. Adding two more rocks and changing no lands does the same, 59% to 66%, and across turns two to six it produces more than twice the extra mana: 0.76 against 0.33. Two rocks match five land swaps on turn five and beat them over a game. The lands are the cheapest fix, not the largest one. They still go first, because a rock has to replace a spell, and a Plains does not.

What each kind costs, as the median of the cycle’s cards at their cheapest printing on November 2, 2026:

The table is the argument for leaving expensive lands until last. For the narrow question of whether a land comes in untapped on the turn you need it, a $0.37 pain land and a $23.45 fetch land get the same answer. Fetch lands do other things that matter in some decks. Whether yours is one of them is a question for after the first pass.

Fill the gaps the reading found

The second stage is the classic advice, applied only where it is true: two-mana rocks if you are short of mana, cheap instant-speed answers if your removal is slow or narrow. On November 2, 2026, Mind Stone was $0.20, Swords to Plowshares $1.25 and Go for the Throat $0.27 at their cheapest printings. If the deck already carries seven or so cheap ramp pieces, as half the sample does, skip the rocks and move on.

Then the plan, then stop

With the mana and the answers in place, what is left of the budget goes to cards that serve the plan. Prefer one that does two things: an answer that also triggers your commander beats either half alone. Then stop and play the batch. You will see about two of the new cards per game, so give it several sessions before you draw conclusions.

How to choose the cuts

Adds are the fun half, but cuts are where the method does its work. Take them in this order.

1. The plan test. Ask of every spell: does it do what the plan sentence describes? If not, is it doing a job the deck is actually short of? A card that fails both goes on the list. A fail is a prompt rather than a verdict, and some cards earn their slot in ways the sentence missed.

2. Surplus before plan. The median deck runs three pieces of ramp and four draw cards over target. If yours does, the weakest of those go before any card that serves the plan. Your fourteenth draw spell is usually worth less than a card that serves the plan.

3. Effects that help the table more than you. “You and target opponent each draw three cards” reads as card draw and counts as card draw on the tile. Read who gets what.

4. Cost without a reason. With a median of 15 spells at five mana or more, some can usually go. But check how each one reaches the battlefield first, because a deck may be built to skip its cost entirely.

5. Not by price, and not because it was reprinted. The median nonland card in these decks had a cheapest printing of $0.31 on November 2, 2026, and 81% of that deck’s spells were under a dollar, so most cuts give up very little. The five most valuable cards held 49% of that value in the median deck, so check what those are before you pull anything. And a reprint is not automatically filler. Only four nonland cards in a median deck appear in ten or more of the 55, and they include Sol Ring (55 of 55) and Arcane Signet (54 of 55). Meanwhile 35 of a median deck’s spells appear in no other deck in the sample, so most of what you are judging is cards made or chosen for that list.

A worked example: Silverquill Influence

• Killian, Decisive Mentor — The commander of Silverquill Influence, from the Secrets of Strixhaven Commander decks released April 24, 2026.

Killian, Decisive Mentor is a three-mana white-black commander with two triggers: whenever an enchantment you control enters, tap up to one target creature and goad it; whenever one or more creatures enchanted by an Aura you control attack, draw a card. The second trigger does not ask whose creature it is. Put Coercive Impetus on an opponent’s creature and it must attack each combat, away from you if it can, and you draw from Killian and from the Aura.

So the plan sentence writes itself: put Auras on creatures, other people’s included, so that enchantments entering keep goading and enchanted attackers keep drawing cards.

The count. The deck has 37 lands, in range. Its seven always-tapped lands and two fetchers make nine slow lands, exactly the median. Draw reads 14 and wipes 3. Ramp reads 4, the lowest in the sample; the tile does not count Land Tax, which fetches basics to hand only while an opponent has more lands. And average mana value is 2.97, third lowest of the 55, with just seven spells at five or more, so the “cut expensive spells” step finds almost nothing here. A curve this low does not need ten ramp pieces. It does want more than four.

The reading. Removal reads 9, in range, and the reading is less kind. Eldrazi Conscription counts because annihilator is an edict of sorts; Screams from Within counts because -1/-1 can kill a small creature. Set those aside and seven remain. Only two of the seven are instant-speed and can hit a creature, Anguished Unmaking and Vanishing Verse, and Verse only hits monocoloured permanents.

The lands. Five swaps take slow lands from nine to four. Path of Ancestry, Silverquill Campus, Forum of Amity, Sunlit Marsh and Terramorphic Expanse go, because their riders do little here. Bojuka Bog and Temple of Silence stay for now, since a graveyard answer and a scry are at least something, and Umbral Expanse can cycle. Turbulent Moor looks slow and mostly is not: it counts the lands of all your opponents together, so three opponents on three lands each is already nine. The spells ask for white 57 times and black 24 times, so the basic going in is a Plains.

The plan test. Does it enter as an enchantment, care about enchantments, or pay off when creatures attack? If not, is it a job the deck is short of? Eight of the 62 spells fail: Keen Duelist, Secret Rendezvous, Nils, Discipline Enforcer, Tomik, Wielder of Law, Inkshield, Mangara, the Diplomat, Shadrix Silverquill and Defacing Duskmage. Three of those stay. Mangara has lifelink and a second trigger that works regardless of goad, Shadrix has flying and double strike, so every point of power an Aura adds hits twice, and Duskmage is a two-mana deathtouch creature that can become an instant draw-two once an opponent draws a second card in a turn. That leaves five, and two more cards pass the test but lose on efficiency. This part is judgment, and another player could reasonably keep Tomik and cut something else.

The seven spell cuts

• Keen Duelist — Counts as draw. Hands one opponent a card every upkeep too.
• Secret Rendezvous — Three cards for you, three for a chosen opponent. Its only link is preparing Defacing Duskmage.
• Nils, Discipline Enforcer — Its counters feed Pearl-Ear, and otherwise it ignores enchantments. One link is not a plan.
• Tomik, Wielder of Law — A fine flyer, but its trigger needs two creatures attacking you, and this deck sends attackers elsewhere.
• Inkshield — Five mana to prevent combat damage in a deck built to redirect combat. At $4.88, the one cut worth keeping in a binder.
• Screams from Within — Passes the test, since it comes back and re-triggers Killian. As removal, it kills X/1s.
• Fallen Ideal — Passes the test too. Three mana for flying and a sacrifice ability this deck has little use for.

Two that look like cuts and are not. Eldrazi Conscription costs eight in a deck with four rocks, but Armored Skyhunter, Herald of Amity and Songbirds' Blessing can each bring it in without paying. And Winds of Rath looks like a self-inflicted wound in a creature deck until you read it: it destroys creatures that are not enchanted, and this deck enchants them on every side of the table.

The twelve swaps. Prices are the add’s cheapest English nonfoil printing on November 2, 2026. Every add was checked in our index for Commander legality and a white, black or colourless identity, and none is a Game Changer.

The cuts give up little: eleven of the twelve had a cheapest printing of $0.33 or less that day, and the twelfth is Inkshield.

What changed.

Note the conditional row: two of the new lands can still enter tapped, a Concealed Courtyard drawn late or a Shattered Sanctum played early. The model counts them as untapped, so its gain is the flattering version.

Open the Silverquill Influence decklist — The full printed list on Manamancer. An account is needed to save decks, and then you can make the swaps and watch the Stats panel move.

What small budgets buy

The same twelve swaps, cheapest jobs first, with four more for anyone who wants to keep going. All figures are from November 2, 2026, and each row includes the rows above it.

That last row takes the wipes tile from three to two on purpose: Promise of Loyalty makes you sacrifice your own enchanted creatures too. It is a case of reading beating the tile, and of the tile being worth knowing while you overrule it.

Two cards that look like obvious adds stayed out. Smothering Tithe is $55.62 and on the Game Changers list. Greater Auramancy is $18.27 and gives your enchanted creatures shroud, which stops you putting your next Aura on them.

The cheap end is not peculiar to this deck. Of the 1,000 most-played nonland cards in Commander by EDHREC rank, 38% had a cheapest printing of a dollar or less on November 2, 2026, 48% of two dollars or less and 67% of five dollars or less, with a median of $2.21. A low price does not make a card right for your deck, but a small budget is rarely what stops a first pass.

What this cannot tell you

Three limits, because the method is narrower than its confident tables.

First, none of this measures winning. We have no record of how these 55 decks perform, upgraded or not, so nothing above promises that twelve swaps will win you more games. What it shows is what the lists contain and what the swaps change.

Second, counts and plan tests are prompts. The counter misreads individual cards, and our plan test on Silverquill Influence is one reading of one deck. If you disagree with a cut after playing the card, you are the one with the better evidence.

Third, a bracket number describes the natural weight class of a shell, not a promise about your list. None of the sixteen adds above is a Game Changer, an extra-turn card or land destruction, but cheap interaction and a smoother mana base still make a deck faster than its printed version. Say so before the game; the Stats panel will show you the bracket reading for your actual list.

What to do next

Pick the deck, write its one sentence, and count. For a guide built around one commander’s engine, our Wretched Ranks upgrade goes deeper on a single list, and the Foundations Commander ranking compares five decks out of the box. If you want a second opinion on the last few slots, the Agent tab can suggest cards checked against the card index, Commander legality and your commander’s colour identity, and you apply or dismiss each one; the guide explains how.

May your Auras always find a creature, and may that creature always attack someone else.