TLDR
To diagnose mtg deck mulligan problems, label why each questionable opening hand fails before changing the list. Separate insufficient mana from missing colors, slow curves, conditional cards, matchup-dependent interaction, and unrealistic keep standards. Test repeated hands from a saved baseline, play out the first three turns, change one coherent part of the deck, and repeat. “Add more lands” is useful only when mana quantity is actually the problem.
A deck that keeps mulliganing is giving you information, but not necessarily the information you think. Two lands and five spells can be excellent in one deck, unplayable in another, and deceptively bad when both lands enter tapped or produce the wrong colors. The useful question is not simply, “How many lands did I draw?” It is, “What prevents this hand from executing the deck’s early plan?”
Start by defining what a keepable hand must do
Write down the minimum job of an opening hand before drawing any test hands. That job should be specific to the deck. An aggressive deck may need an untapped source and a turn-one threat. A control deck may prioritize stable color access and interaction. A Commander engine deck may need early development that leads into its commander rather than a hand full of impressive six-mana payoffs.
Under the June 19, 2026 Comprehensive Rules, players normally begin with seven cards. After taking a mulligan, a player redraws and then puts a number of cards equal to the mulligans taken on the bottom of the library. Multiplayer games and Brawl do not count the first mulligan toward that bottoming requirement. The complete procedure is available in the official Magic Comprehensive Rules.
That procedure makes mulligan decisions consequential, but it does not create one universal keep rule. Wizards’ strategy guidance notes that deck type, format, opponent, early plays, and matchup-specific cards all affect the decision. Even a familiar guideline such as keeping hands with two to five lands is only a starting heuristic. For general in-game keep-or-ship principles, see the MTG mulligan guide. This guide focuses on the different problem: repairing a list that repeatedly offers bad choices.
Use six labels for MTG deck mulligan problems
Give every rejected or suspicious hand one primary failure label. A hand can have several weaknesses, but choosing the dominant one makes the eventual deck change easier to interpret.
| Failure label | What the hand looks like | What to inspect |
|---|---|---|
| Mana quantity | Too few usable sources, too many sources, or ramp that arrives too late | Land count, mana-producing cards, draw and mana sinks |
| Color access | Enough lands, but not the colors or untapped sources needed | Colored pips, source distribution, tapped lands and splash costs |
| Curve and timing | Cards are castable eventually, but the deck has no meaningful early turn | Mana curve, first plays, expensive congestion and sequencing |
| Conditional cards | The hand contains payoffs without enablers, targets or complementary pieces | Package size, redundancy, standalone utility and role balance |
| Matchup dependence | Reactive cards may be excellent or nearly blank depending on the opponent | Game-one assumptions, sideboard plan, play versus draw |
| Keep criteria | The hand can function, but you reject it—or it cannot function, but you keep it | Written keep standards, required early actions and wishful thinking |
1. Check mana quantity without blaming every bad hand on land count
A genuine mana-quantity problem appears when otherwise functional hands repeatedly have too few usable mana sources to reach the deck’s important turns. The opposite problem also exists: hands with plenty of lands but too little action, especially when the list has limited card selection or few productive ways to spend excess mana.
Start with the turns that matter. How much mana does the deck need on turns one, two, and three? If its first meaningful play costs four, determine whether the earlier turns develop mana, select cards, or interact. A two-land hand with cheap selection may function differently from a two-land hand whose “ramp” also costs three.
Do not count every nominal mana card as though it serves the same role. Expensive ramp cannot rescue an opener that cannot cast it. Conditional mana may depend on another permanent. A land that always enters tapped cannot satisfy an early untapped-source requirement. Use a deck builder to inspect mana roles rather than viewing the land total in isolation; the castability-focused deck builder guide explains how to connect mana structure to the spells that must be cast on time.
2. Separate land count from colored-source problems
A three-land hand can still be a mana failure. If those lands cannot produce the required colors in the required order, adding another random land may not repair anything.
For each questionable opener, state the actual sequence. Instead of saying, “I have three lands,” say, “Turn one tapped land, turn two blue source, turn three another tapped land; my double-white spell remains uncastable.” That sentence identifies color and timing as the bottleneck.
Inspect four things separately: total mana sources, sources of each needed color, sources that enter untapped by the relevant turn, and the intensity of colored costs. Splash cards and double- or triple-pip spells create different demands. A mana base that produces every color eventually may still fail a deck that needs one color immediately and another twice on turn three.
Changing the mana base may therefore mean adjusting color distribution, replacing tapped lands, reducing demanding costs, or removing a splash—not merely increasing the land total. This is why a curve view and a deck list should be used together; deck builders and mana-curve analyzers answer different questions.
3. Test whether the curve creates empty early turns
Some hands have lands and spells yet accomplish nothing. That is usually a timing problem. The cards may be individually castable, but too many occupy the same expensive slot or depend on setup that begins too late.
Record the first meaningful play in each hand. Then distinguish a real early play from a card that merely has a low mana value. A cheap protection spell with nothing to protect is not development. Removal with no relevant target may not be an action. A one-mana combo piece that does nothing until turn five may be setup, but only if the rest of the plan can support that tempo cost.
A mana-curve analyzer can expose congestion, but the shape needs interpretation. Separate ramp, card selection, interaction, threats, and setup pieces within each cost band. If opening hands repeatedly contain multiple expensive cards and no bridge to them, use the expensive-card congestion workflow to decide whether the problem is the payoffs, the ramp, or the missing early plays.
4. Find narrow cards and the wrong halves of packages
Synergy decks can generate hands that look powerful in the deck builder but fail as seven-card combinations. Common examples include multiple payoffs with no enabler, tutors without worthwhile early targets, sacrifice rewards without expendable permanents, or reactive spells that answer only a narrow class of threats.
Tag cards by functional role, then look for packages that need another category before they do anything. Ask whether each piece has acceptable standalone utility and whether the deck contains enough interchangeable enablers. Redundancy helps only when the redundant cards perform the needed role; drawing three versions of the same payoff is not the same as drawing a complete engine.
When this label appears often, do not automatically remove the deck’s signature synergy. First consider replacing the weakest narrow card with a flexible piece that advances more than one role: an enabler that also selects cards, interaction that also develops the board, or a payoff that remains useful without the ideal setup.
5. Account for matchup, sideboarding, and play versus draw
A hand full of narrow interaction may be excellent when the opponent is known and poor in game one against an unknown field. Likewise, a slower hand may be acceptable on the play but unable to recover on the draw. Log that context instead of treating the two decisions as identical.
For post-sideboard testing, specify the opponent’s plan and what your deck must answer. If the keep depends on one reactive card being relevant, say so. Testing against a representative lineup is better than drawing context-free hands when the question is matchup performance; a Constructed testing gauntlet can help separate broadly weak openers from matchup-specific ones.
6. Audit your keep criteria
Sometimes the list is reasonable and the decision rule is broken. Players keep hands because they contain a favorite payoff, because the land count falls within a familiar range, or because one future draw would make everything work. They also throw back functional sevens while searching for an ideal curve that the deck rarely produces.
Write keep standards as actions, not card counts. For example: “This hand can cast an early play, produce the deck’s primary color, and reach its key turn with one ordinary draw.” Then write automatic warning signs, such as needing two specific draws, having no untapped source before turn three, or holding only payoffs.
Do not demand the same opener from every strategy. A resilient value deck can keep a merely functional hand more readily than a linear deck whose opener must assemble a particular sequence. The standard should reflect what the deck is built to do, not what a generic mulligan slogan says it ought to do.
A repeatable opening-hand testing workflow
Starting-hand testing works best as a controlled deckbuilding loop rather than a dramatic reaction to one ugly seven. Goldfishing can reveal mana access, sequencing, and how quickly an unopposed plan develops, but it does not test resistance to interaction, combat pressure, sideboarding, or multiplayer politics.
- Save the current deck as a baseline before making changes.
- Define the deck’s early plan and write minimum keep criteria.
- Draw repeated opening hands using the same randomization method.
- For every hand, record play or draw, known matchup context, and keep or mulligan.
- State the intended turns one through three using the lands and spells actually present.
- Assign one primary failure label: mana quantity, color access, curve, conditional cards, matchup, or keep criteria.
- Goldfish kept hands far enough to see whether the promised sequence works without ideal draws.
- Count recurring labels rather than treating every rejected hand as the same problem.
- Change one coherent variable, such as two expensive cards, a color-source package, or a group of narrow payoffs.
- Repeat under the same conditions, then confirm promising changes in real games.
There is no universal number of sample hands that proves a deck is fixed. A small batch can reveal obvious structural issues; subtle or matchup-dependent questions need broader testing. The important part is consistency: use the same baseline, conditions, labels, and keep standards so the comparison means something.
Deckbuilder and starting-hand testing checklist
- Confirm the format, deck size, primary plan, and key turns.
- Tag lands, ramp, selection, interaction, enablers, payoffs, threats, and flexible cards.
- Review total mana separately from colored and untapped sources.
- Inspect the curve by functional role, not only mana value.
- Write what a minimum keep must accomplish by turns one through three.
- Log play versus draw and whether the opponent is known.
- Label each failed hand with one dominant cause.
- Check whether conditional pieces have enough useful partners.
- Save the baseline and change only one coherent package at a time.
- Use goldfishing for internal sequencing and real games for matchup, interaction, and sideboard questions.
Fix the recurring failure, not the latest bad hand
The best response to frequent mulligans is not a universal land number or a looser keep rule. It is a diagnosis. Identify whether the deck lacks mana quantity, the right colors, untapped access, early actions, balanced synergy pieces, broadly relevant cards, or realistic keep standards. Then make the smallest change aimed at that recurring failure.
Start by saving the list, writing down what a functional opener must do, and testing hands without negotiating with yourself after seeing the cards. If the same label keeps appearing, you have a useful deckbuilding target. If the labels are scattered, the deck may be experiencing ordinary variance—or trying to execute too many different plans at once.