clinic · an initiative without a goal

An initiative without a goal

The Initiative Brief was signed. The first cycle ran. The signal didn't move; the cycle 2 plan was already in motion. The signal didn't move again. The portfolio review opened the brief looking for the kill criterion — and found, in its place, a paragraph that began "we'll review progress quarterly". By cycle five, the initiative had consumed 40% of the team's capacity for nine months. No one could remember exactly when stopping had stopped being an option.

The artefact

Excerpt — Initiative Brief, "Grader Workflow Modernisation", September 2025

Population: Teaching faculty across our LMS deployments.

Journey: A grading week.

The bet: Modernising the grader workflow will reduce friction across the grading process and improve grader satisfaction, leading to higher overall throughput and a stronger competitive position in the LMS market.

Biggest unvalidated assumption: That the current workflow has identifiable points of friction that can be removed.

Success signal: Across the initiative, we expect to see grader-experience improvements and increased grading throughput.

First Feature Brief: /briefs/feature/grader-ui-refresh.md

Kill criterion: We will review progress quarterly. If at any point the initiative is not delivering value, we will reassess.

Sign-off: [PO] · [Leadership]

The brief was signed. Each of the seven sections was filled. The discipline of writing the Initiative Brief was technically completed.

Nine months later:

• The initiative had run five cycles. Each cycle shipped something — a UI refresh, a notification refactor, a queue redesign, a feedback templating system, an inline preview.
• None of the cycles had produced a measurable improvement in any specific signal. The team had been busy.
• Three different cycle predictions had been missed; each was explained away (sample too small, cohort wasn't representative, the metric was the wrong proxy).
• The 2026-Q2 portfolio review opened the Initiative Brief looking for the kill criterion. The line "we will review progress quarterly" turned out to be unkillable.
• The leader running the review said: "I think we need to reassess this initiative." The PO said: "We've made real progress — the last cycle's queue redesign got positive feedback." Nobody could quite point to the line in the brief that said stop.

The initiative continued for another quarter.

What's wrong?

Stop. Find three things wrong before reading the diagnosis.

Diagnosis (open when ready)

1. The bet is generic, not specific

Modernising the grader workflow will reduce friction across the grading process and improve grader satisfaction, leading to higher overall throughput and a stronger competitive position in the LMS market.

This is not a bet. It is a hope dressed as a sentence. There is no mechanism — no because. There is no specific population — teaching faculty across our LMS deployments is a category, not named persons. The bet cannot be falsified, because there is no specific if-then that reality could disagree with.

The corpus rule from Practice · Writing Initiative Briefs · Step 3: the bet is three sentences. If [intervention] for [population], then [signal] will move because [mechanism].

A real bet for this initiative might have been:

If we remove the Hebrew-name lookup workaround for graders at the flagship campus, then median focused-grading time will fall from 47 minutes to under 15, because removing the alt-tab interruption restores uninterrupted attention long enough to grade in flow. We are running this bet because the next 100-grader deployment depends on per-cycle grading being a 20-minute task, not a 45-minute one.

That bet is specific, mechanism-named, and falsifiable. The cycle 1 check either moves the signal or it doesn't — and if it doesn't, the team learns that the mechanism was wrong, not just that things didn't go well.

The actual brief's bet — "reduce friction... improve satisfaction... higher throughput... stronger competitive position" — cannot be refuted. Every cycle can claim some friction was reduced, some satisfaction improved, etc. The bet is unkillable because it doesn't say enough to be wrong.

2. The biggest unvalidated assumption is tautological

That the current workflow has identifiable points of friction that can be removed.

This says: we assume there is something to fix. That isn't an assumption; that's the premise of doing the work. A real biggest-unvalidated-assumption surfaces the thing that, if false, kills the bet:

Graders skip afternoon grading batches because of fatigue from morning friction, not because afternoon is when they teach.

If the assumption is false, the bet is wrong — graders won't grade more in the afternoon even with a faster morning flow, because they're not free in the afternoon. The corpus's discipline is that the assumption is the thing that, if refuted by the first cycle's check, kills the initiative.

A tautological assumption is unfalsifiable. The first cycle ran; nothing was refuted; the initiative continued because nothing could refute it.

3. The kill criterion is the failure shape itself

We will review progress quarterly. If at any point the initiative is not delivering value, we will reassess.

This is not a kill criterion. It is a paragraph that describes the kill criterion's failure shape. There is no signal, no threshold, no date. Every word is renegotiable:

• Review progress quarterly — review by whom, against what?
• Not delivering value — value as defined by whom?
• Reassess — what does reassess produce? Continuation, by default.

The corpus rule from Practice · Kill criterion: the criterion is one sentence with a specific signal, a specific threshold, and a specific date.

A real kill criterion for this initiative:

If by 2026-03-01, median focused-grading time across observed grading cycles at the flagship campus has not fallen below 25 minutes (from the 47-minute baseline), the initiative is killed and the team is freed for next-quarter work. The bet's mechanism — focused-time recovery via workaround removal — is considered refuted.

That criterion fires automatically. The signal either crosses the threshold by the date or it doesn't. The conversation at the review takes thirty seconds, not thirty minutes.

The actual brief's "criterion" required a judgement — and a team that had spent nine months on the initiative was no longer the right judge. The brief should have been signed by a leader who anticipated this exact moment — and wrote the criterion to fire automatically.

4. (Bonus) The success signal is also unfalsifiable

Across the initiative, we expect to see grader-experience improvements and increased grading throughput.

Improvements without numbers; increased without a baseline. Same failure shape as the bet. Each cycle could point to some improvement and the success signal stayed alive.

A real success signal:

Cycle 1: median focused-grading time falls from 47 to <20 min.Cycle 2: at least 1 in 3 graders adds an afternoon batch they currently skip.Cycle 3: weekly grading completion rises from 62% to >80% at the flagship campus.

Three signals, three numbers, three cycles. The first that fails refutes the bet's next-cycle mechanism.

The fix

# Initiative Brief — Hebrew-grading workflow (rewrite)

## Population
Teaching faculty at our flagship campus (Acme University,
CS department). 28 active graders, all teaching CS101 or
CS201 in 2026-Q2.

  Representative named persons:
  - Gal — TAU CS101, grades 2 mornings/week
  - Dina — Hebrew U CS-stats, grades evenings
  - Yael — flagship campus CS101 TA, observed 2026-04-22

## Journey
A grading week (J1–J8 mapped). In scope: J2 (open queue),
J3 (open submission), J5 (type feedback). Out of scope:
J4 (read/assess), J6 (submit), J7 (return), J8 (weekly
totals).

## The bet
If we remove the Hebrew-name lookup workaround for graders
at the flagship campus, then median focused-grading time
will fall from 47 min to under 15, because removing the
alt-tab interruption restores uninterrupted attention long
enough to grade in flow. This matters this quarter because
the next 100-grader deployment depends on per-cycle grading
being a 20-min task, not a 45-min one.

## Biggest unvalidated assumption
Graders skip afternoon grading batches because of fatigue
from morning friction, not because afternoon is when they
teach. If teaching-time is the real cause, this initiative
ships a faster morning flow that no one uses in the
afternoon — and the bet's mechanism is refuted.

## Success signal (across 3 cycles)
Cycle 1 (Q2): median focused-grading time falls 47 → <20 min.
Cycle 2 (Q3): ≥1 in 3 graders adds an afternoon batch.
Cycle 3 (Q4): weekly grading completion rises 62% → >80%.

## First Feature Brief
/briefs/feature/2026-q2-hebrew-name-grading.md

## Kill criterion
If by 2026-09-01, median focused-grading time across
observed grading cycles at the flagship campus has not
fallen below 25 min (from the 47-min baseline), the
initiative is killed and the team is freed for
next-quarter work. The bet's mechanism — focused-time
recovery via workaround removal — is considered refuted.

## Sign-off
PO:         Alex (PO) · 2026-04-15
Leadership: [Leadership name] · 2026-04-15

Every section is specific, dated, and kill-able. The first cycle's check fires on a real date against a real number. If the signal doesn't move below the kill threshold by 2026-09-01, the initiative is killed — automatically, without judgement.

The conversation at the 2026-Q3 portfolio review takes 30 seconds: Signal at 13.2 min. Criterion met with margin. KEPT.

Or, if the bet had been wrong: Signal at 41 min. Criterion missed (>25 min threshold). KILLED. Team rotates to [next priority] at the start of next cycle.

Where this comes from in the chain

This failure traces to Strategy (Level 1). The Initiative Brief is the chain's strategic artefact; when it is signed without discipline, every level below it inherits the looseness. Cycle 1 ships against a generic bet. Cycle 2 doesn't know what to refute. The portfolio review can't kill what was never kill-able.

The structural fix is at L1 — the Initiative Brief's gate is can you write a kill criterion? If not, the initiative is not yet ready to be signed. The PO and Leadership refuse to sign until the criterion is sharp.

A senior leader catches this in 60 seconds: open the brief, find the kill criterion, ask would this fire automatically on a specific date? If not, send the brief back.

The deepest learning: an initiative whose Brief cannot be killed cannot be honestly led. The kill criterion is what protects the team — from working on the wrong thing, from the sunk-cost trap, from the unspoken pressure of we've already invested so much. The criterion is the discipline that says the bet either holds or it doesn't, and the team is freed regardless of which.

See also

• Practice — Writing Initiative Briefs · Kill criterion · Quarterly portfolio review
• Template — Initiative Brief — the kill criterion section
• Checklist — Portfolio review · agenda — reads kill criteria aloud
• Canon — Did We Serve · The Portfolio · When to stop · Why We Build · Initiative Identification
• Principle — Predictions over plans