//pragmatic leaders

MARK

Good judgment is choosing what to do when the answer is not obvious.

Product work rarely gives you a clean answer. The data conflicts. Customers want different things. The loudest stakeholder may be wrong. Time and money are limited. MARK is a structure for seeing how well you make the call anyway, and explain why.

Map your MARK →Open Your MARK →

what MARK does for you

Build a credible story of how you judge.

Your MARK becomes a source-linked profile of the judgment your work demonstrates, where the evidence is still thin, and what would be useful to practise next.
  1. 01

    See what your work already shows.

    Every claim links back to the answer or activity that produced it. You can inspect, challenge, or retract the read.

  2. 02

    Know what to practise next.

    MARK separates a real development edge from a criterion that is simply untested, so your next attempt has a reason.

  3. 03

    Build a story others can inspect.

    Over time, Your MARK becomes a source-linked record you can use in a 1:1, interview, promotion case, or team review.

//PL uses the MARK symbol wherever a judgment call can add to this record.

the answer, in one picture

Good judgment rests on four pillars. Each pillar has three criteria.

Map, Acuity, Resolve, and Know-how form MARK. The twelve criteria make each pillar specific enough to observe in a real answer, decision, or conversation. The radar shows the quality demonstrated so far; it is a changing shape, not a permanent score.

one possible MARK

An illustrative shape, not a target score
WorthKillHaltSignalReframeBetPowerHoldMissRoomUserTasteMMapAAcuityRResolveKKnow-how
  1. 1Developing
  2. 2Competent
  3. 3Proficient
  4. 4Expert
M

Map

Choose what matters

  • WorthDeciding what’s worth building among everything you could ship.
  • KillKilling your own idea when sunk cost says don’t.
  • HaltDeciding when not to build at all.
A

Acuity

See what is really going on

  • SignalReading conflicting evidence to a decision.
  • ReframeSpotting and fixing the wrong question before solving it.
  • BetSizing investment under uncertainty (incl. reversibility / one-way doors).
R

Resolve

Make and own the call

  • PowerSaying no to power — telling the exec the call is wrong.
  • HoldHolding a hard call without folding or bluffing.
  • MissOwning a clean miss without deflection.
K

Know-how

Read people, users, and quality

  • RoomReading the room — what people actually mean.
  • UserReading user/customer truth — who you’re really building for.
  • TasteJudging quality you didn’t make (incl. AI output).

Each criterion has a precise definition and four behavioral levels. See all 48 behavioral examples ↓

what the shape can show

MARK exposes the pattern behind “good” or “bad” judgment.

A single average would hide the useful part. The shape shows where judgment is strong, uneven, absent, or still untested, and which imbalance could become consequential.

sound, but uneven

A person can show strong judgment without being equally strong everywhere.

This shape is strongest on Reframe and User. It is less tested on stopping work and speaking against power. The useful question is what to practise next, not whether the polygon looks large.

a consequential blind spot

You can execute the wrong work very well.

Here Acuity, Resolve, and Know-how look strong, but Map is weak. The person may reason well after accepting the wrong agenda. MARK makes that imbalance visible.

MARK is not a seniority ladder. A title from L1 to L7 changes the scope and stakes of the work, not the scoring rules. A senior title does not automatically produce a stronger MARK; evidence from judgment calls does.

how Your MARK grows on PL

Your MARK grows one Trace at a time.

PL can only learn from work it can actually see. A Trace is one source-linked observation from one activity, plus outcomes or human reviews you deliberately add later.

how to read one Trace

Every criterion is read on two separate axes.

Quality asks: How strong was the judgment in this piece of work? Confidence asks: How much should we trust that this judgment will repeat?

the complete system

MARK separates what you did, how well you judged, and how much the evidence deserves to be trusted.

01 · work PL can seeOne answer or activityManual, Practice, Cases, an action plan, or a starting self-view.
02 · criterionOne of the twelve is visiblePL links the interpretation to the exact response that showed it.
03 · traceOne sourced observationA Trace belongs to this piece of work. It is not yet a lasting claim.

quality · how well?

How strong was the judgment in this work?

  1. L1
    DevelopingDemonstrated with guidance.
  2. L2
    CompetentDemonstrated independently in familiar situations.
  3. L3
    ProficientDemonstrated in complex situations; helps others reason.
  4. L4
    ExpertChanges how a team makes this kind of call.

confidence · how sure?

How much should we trust that it will repeat?

  1. 01
    Self-reportedA starting hypothesis that PL has not observed.
  2. 02
    ObservedOne sourced activity shows the criterion.
  3. 03
    RepeatedSeparate work shows it in different situations.
  4. 04
    VerifiedOutcomes or credible reviewers support the interpretation.
one observationTrace
at least two pieces of workPattern
at least three pieces, two contexts, high quality, supported outcomesHallmark

where the work comes from

Use PL, then let separate activities and outcomes strengthen or challenge the read.

starting hypothesisMap your starting pointA self-view that tells PL what to test. Useful, but never treated as proof.try it →observed evidenceAnswer a scored Manual exercisePL can read the reasoning in the answer you submit.try it →observed evidencePractise a hard conversationPL can read the judgment shown across the conversation.try it →observed evidenceMake a call in a CasePL can compare the choice, reasoning, and trade-off you made.try it →observed evidenceBuild an action planPL can read how you turn a development edge into a concrete next step.try it →follow-up evidenceAdd what happened nextAn outcome can confirm, challenge, or overturn PL’s earlier interpretation.add it →

What PL does not do: it does not watch your Slack, WhatsApp, calls, or meetings. Work outside PL enters Your MARK only when you choose to add an outcome or ask a person to review it.

the complete standard

48 behavioral examples make the twelve criteria testable.

Each cell describes what one criterion looks like at one quality level. The matrix is the detailed reference; the radar and system diagram are the mental model for reading it.

The complete MARK standard

12 criteria × 4 quality levels = 48 behavioral examples. Grouped by the four MARK pillars. Click any cell to open that criterion at that level.

judgment criterionL1DevelopingL2CompetentL3ProficientL4Expert
MMap· direction: what’s worth building
WorthDeciding what’s worth building among everything you could ship.builds whatever was requested most recently, without checking whether it moves a metric or fills a real gap.ranks options by user impact and feasibility on familiar problems, but stalls or seeks input when tradeoffs involve unfamiliar constraints or high business stakes.spots the trap in a promising idea before the team commits — flags why a high-demand feature solves the wrong problem, and redirects scoping without waiting to be asked.reframes how the org defines worth — replaces inherited criteria with ones built from evidence, so future prioritisation decisions change, not just the current one.
KillKilling your own idea when sunk cost says don’t.waits for a manager to flag a project should stop, then agrees only after repeated prompting.kills ideas with clear evidence of failure, but hesitates when investment is high or the signal is mixed — acts after a peer or manager names the problem aloud.spots warning signs early, names the kill decision aloud, and explains the reasoning so others can apply the same judgment independently.names the sunk cost trap aloud, cuts publicly, and becomes the person others bring their hardest stops to — building the team's habit of killing bad ideas early.
HaltDeciding when not to build at all.flags "we shouldn't build this" signals only after a senior teammate names them first; does not identify halt conditions independently.flags low-demand or duplicative build requests and recommends stopping; defers when the business case is ambiguous or a stakeholder pushes back.spots unviability conditions before work begins and walks teammates through the reasoning so they can make the call independently next time.kills work others won't touch by naming the real cost of building it, then reframes the team's default from "how do we build this?" to "should we?"
AAcuity· discernment: reasoning under ambiguity
SignalReading conflicting evidence to a decision.flags contradictions only after a teammate names them, then defers to whoever spoke last instead of weighing the evidence.sorts conflicting signals on familiar problems and names a lead interpretation; stalls or seeks input when evidence patterns fall outside past experience.spots conflicting signals as noise before others flag them, then names the specific failure mode that would invalidate the leading interpretation.reframes which signals matter before others form the question, then calls the decision — setting the standard the team uses on the next hard call.
ReframeSpotting and fixing the wrong question before solving it.accepts the problem as stated and solves it; questions the framing only after a manager or peer explicitly flags that the brief is wrong.catches misframed problems in familiar territory; raises the reframe only after someone else signals doubt on high-stakes or unfamiliar work.catches misframed problems before work begins, redirects the team to the right question, and documents the reframe so others learn the pattern.reframes the problem before the room commits to the wrong one — and the team adopts that framing as the new starting point.
BetSizing investment under uncertainty (incl. reversibility / one-way doors).sizes bets by gut or copying past projects; skips reversibility checks. when pressed to justify commitment level, needs a manager to reframe or reduce the ask.sizes bets on familiar decisions by checking reversibility and adjusting spend, but seeks sign-off when stakes rise or the situation falls outside past experience.sizes bets on novel decisions using explicit reversibility checks, names the one-way doors before commitment, and adjusts investment based on what new evidence would change the call.reframes what counts as a good bet — teaches the team to size commitments against reversibility, not confidence, so the org builds fewer expensive traps.
RResolve· conviction: holding a position under pressure
PowerSaying no to power — telling the exec the call is wrong.agrees with the exec in the room, then surfaces concerns only in private or after the call.names a disagreement with the exec when evidence is clear; backs down when the exec pushes back without new argument.holds the position under direct pushback from senior stakeholders, names the cost of the wrong call in concrete terms, and does it in the room — not in a follow-up note.reframes the stakes before the exec decides, not after — making the cost of the wrong call visible in terms leadership can't dismiss.
HoldHolding a hard call without folding or bluffing.folds under the first credible challenge — rephrases the position as a hypothesis or defers to the objector.holds a stated position under routine pushback; escalates to a manager before standing firm when stakes rise or a senior stakeholder objects.holds a hard call under sustained pushback from multiple stakeholders, names what evidence would change the view, and stays on the position in the room — not in the follow-up note.holds the call when everyone else has folded, names the failure mode that will happen if the team pivots, and absorbs the social cost of being right publicly.
MissOwning a clean miss without deflection.admits a miss only when pressed, then adds context that shifts blame before the acknowledgment lands.admits the miss directly when prompted, names what went wrong without blaming data gaps or timing, but waits for a postmortem or 1:1 rather than flagging it in the moment.spots recurring miss conditions in a new situation and flags them before the outcome repeats. coaches teammates to name their misses directly, without softening the cause.names the miss publicly, explains what broke in the thinking, and reframes how the team should evaluate similar calls going forward.
KKnow-how· the read: human + quality judgment
RoomReading the room — what people actually mean.takes meeting statements at face value; misses subtext and unspoken objections that peers read without prompting.picks up on what stakeholders mean in routine conversations; asks for clarification before acting when stakes rise or signals conflict.reads what is not said — names the real objection in the room before the speaker does, then addresses it directly.calls out the unspoken trade everyone is dancing around, then forces a decision the team has been avoiding — and the room moves because of it.
UserReading user/customer truth — who you’re really building for.builds for the loudest or most visible user segment without checking who actually uses the product or what they actually do.conducts user interviews and maps patterns to a core user profile, but defaults to familiar user types and needs prompting to question assumptions when the problem space shifts.reframes who the actual user is when the team has assumed the wrong person, then tests that reframe against behavioral evidence before committing.reframes who the real user is when the team has drifted; that reframe changes what gets built.
TasteJudging quality you didn’t make (incl. AI output).ships the first thing that works; can't tell whether the output is good or just done, so waits for a teammate to point it out.catches obvious misses against the brief, but freezes on calls like 'is this copy on-brand?' or 'is this visual confusing?' when there's no rubric to lean on.names what makes a piece of work fail before it ships — including ai-generated output — and explains the criteria so others can apply the judgment independently.sets the criteria others use to judge quality — names what makes a piece of work succeed or fail before the team can articulate it, and those calls prove right.
MMap

direction: what’s worth building

AAcuity

discernment: reasoning under ambiguity

RResolve

conviction: holding a position under pressure

KKnow-how

the read: human + quality judgment

reference and portability

Inspect the standard. Carry it elsewhere.

See what changed, compare what MARK does and does not measure, or download all 48 behavioral examples.
MARK comparedWhere MARK stops and career ladders or operating models begin.compare →Version historyWhat changed, and how historical results remain interpretable.read →Download CSVAll 48 anchors for systems and calibration tooling.download →Download ExcelThe formatted standard for Excel, Sheets, or Numbers.download →

start with a hypothesis

Map your MARK. Then let your work argue back.

The self-view gives you a starting read. Scored Manual exercises, Practice, Cases, action-plan submissions, and outcomes you add later can confirm or overturn it.Map your MARK →
pragmatic leaders · PL Standard v3.2 · panel-upheld
4 pillars · 12 criteria · 4 quality levels · confidence linked to evidence