PM Cheat Sheet: Prioritization, Delivery & Control

Quick start

If you only do these 5 things, you’ll already be better than most “PM by title”.

30-minute kickoff checklist

Goal (Outcome): what changes for the customer/business?
Success metrics: how will we know it worked? (numbers, dates, quality)
Scope: what is in / out? (write it down)
Stakeholders: who cares, who decides, who executes?
Constraints: time, budget, tech, legal, dependencies
Risks: top 3 things that could kill the project
Next 7 days: first deliverable + owners + dates

Minimum viable PM artifacts

One-page charter: goal, scope, timeline, owners, constraints
Backlog / task list: prioritized, owned, sized (roughly)
Milestones: “what by when” (don’t worship Gantt)
RAID log: risks, assumptions, issues, decisions
Weekly status: 6 lines (template below)

Tip: If you can’t keep these updated weekly, you’re managing “hope”, not a project.

The golden triangle (how to say “no” nicely)

When one changes, at least one other changes:

Scope / Quality
Time
Cost / People

“We can add X. Which do you prefer: push the date, add budget, or drop Y?”

Meeting rule (simple, powerful)

Start with Purpose + Agenda + Decision points.
End with Decisions + Actions (Owner, Due date).

If it has no decision or action, it’s probably not a meeting.

Which tool when?

Use this to choose the right lens in under 30 seconds.

Fast picker

Many ideas + many voices → KJ method (grouping + voting)
Need to pick top items fast → ICE / RICE
Need economic sequencing across teams → WSJF
Scope negotiation for a release → MoSCoW
Feature satisfaction trade-offs → Kano
Where value leaks in the journey → AARRR
Reduce surprises & manage risk → RAID

Weekly rhythm (works almost everywhere)

Daily (10–15m): done / next / blocked
Weekly (30–45m): plan vs actual + risks + decisions
Bi-weekly/monthly: stakeholder review + priorities check

If stakeholders don’t make decisions, Agile turns into “chaos with ceremonies”.

Prioritization

Scoring is a tool for conversation, not a machine that produces truth.

MoSCoW (scope negotiation)

Use when: stakeholders want “everything”, you need a clear release boundary.

Must: non-negotiable for the release to work
Should: high value, but can ship without
Could: nice to have
Won’t (this time): explicitly not now

Rule: “Must” items should fit into ~60% of capacity. If everything is Must → nothing is.

ICE (fast & dirty)

Use when: you need quick ranking and data is limited.

ICE = Impact × Confidence × Ease

Impact: 1–10 (how much it moves the needle)
Confidence: 0.5 / 0.8 / 1.0 (how sure you are)
Ease: 1–10 (how easy to implement)

Pitfall: “Ease” becomes fantasy without engineering input.

RICE (better when you can estimate reach)

Use when: you can estimate how many users/accounts it touches.

RICE = (Reach × Impact × Confidence) ÷ Effort

Reach: users per period (e.g., per month)
Impact: 0.25 / 0.5 / 1 / 2 / 3 (tiny → massive)
Confidence: 50% / 80% / 100%
Effort: person-weeks (or story points)

Tip: use ranges (min/likely/max) instead of fake precision.

WSJF (economic sequencing)

Use when: multiple teams, limited capacity, “what first?” debates.

WSJF = Cost of Delay ÷ Job Size Cost of Delay = User/Business Value + Time Criticality + Risk Reduction / Opportunity Enablement

Score each CoD component 1–10
Job Size: 1–13 (relative size is fine)

WSJF helps ordering. It doesn’t tell you if the whole portfolio makes sense.

AARRR funnel (problem location)

AARRR doesn’t give a score. It tells you where your value is leaking.

AARRR (Acquisition → Activation → Retention → Revenue → Referral)

Acquisition: people discover you (traffic, CPC, SEO, referrals)
Activation: first meaningful moment (signup-to-value)
Retention: they come back (D7/D30 retention, churn)
Revenue: they pay (conversion to paid, ARPU, LTV)
Referral: they bring others (invite rate, virality, NPS driver)

How to use: 1) Pick ONE stage to improve. 2) Define ONE metric + baseline. 3) Generate ideas for that stage → score via ICE/RICE → run small bets.

Common mistake

“We need more features.”

Usually you need to improve a specific stage (Activation or Retention) before adding more stuff.

KJ method (affinity mapping + voting)

Best when everyone has ideas and alignment is missing.

KJ in 5 steps

Silent idea write (5–10 min)
Group into themes (no debating)
Name clusters (what problem is this?)
Dot-vote
Turn winners into backlog candidates + owners

Pitfall: debating too early. KJ works because it’s silent-first.

Turn KJ output into action

Cluster → Problem statement → Success metric → 2–3 solution ideas → ICE/RICE → pick 1 bet

KJ gives alignment. Scoring gives sequencing. Experiments give learning.

Kano (feature satisfaction)

Use this to stop building “cool” features while basics are broken.

Kano categories

Must-be: missing → angry; present → neutral
Performance: more = better (linear)
Delighters: unexpected, create excitement
Indifferent: doesn’t matter
Reverse: some users hate it

Quick rule

Cover Must-be first → then Performance → then Delighters.

If your support inbox is on fire, do not chase Delighters.

Delivery modes (Waterfall / Scrum / Hybrid)

Stop the religious wars. Pick what fits uncertainty and constraints.

Waterfall (classic)

Best when: requirements stable, compliance heavy, high cost of change.
Strength: predictability, documentation.
Risk: late discovery of “wrong thing”.

Scrum (agile)

Best when: uncertainty high, feedback needed, product evolving.
Strength: learning loops, adaptability.
Risk: ceremonies without decisions.

Hybrid

Best when: fixed milestones + iterative build inside.
Example: fixed go-live date + sprints; stage gates + agile execution.

Rule of thumb

Uncertainty high → shorten cycles, increase feedback.
Dependencies high → strengthen interfaces, planning, and ownership.
Compliance high → add gates & documentation, but keep iteration where possible.

Control layer (RAID + stakeholders + RACI)

These reduce surprises and politics pain.

RAID log

Risk: might happen → probability + impact + mitigation
Assumption: believed true → verification date / trigger
Issue: happening now → owner + next action + ETA
Decision: agreed choice → date + rationale + owner

Stakeholders (Power × Interest)

High power + high interest: manage closely (weekly touch)
High power + low interest: keep satisfied (short updates)
Low power + high interest: keep informed (notes/summary)
Low power + low interest: monitor

Most PM problems are stakeholder problems wearing a “process” mask.

RACI (use only when confusion appears)

R Responsible: does the work
A Accountable: owns outcome (1 person max)
C Consulted: gives input
I Informed: needs updates

Rule: If you have 3 Accountables, you have 0.

Change control (mini)

Before saying yes:

What’s the value?
What’s the impact on time/cost/quality?
What will we drop or delay instead?

Confirm in writing: “Approved change = new scope + new dates.”

Copy/paste templates

If you use these weekly, you will look “senior” without pretending.

Weekly status (6 lines)

Project: … | Period: … Overall: GREEN / AMBER / RED (why in 1 sentence) Done: 2–5 bullets Next: 2–5 bullets Risks: 1–3 bullets + mitigation Asks / Decisions needed: 1–3 bullets (owner + due date) Next milestone: date + confidence (High/Med/Low)

Escalation message (options, not drama)

Context: … Problem: we’re blocked by … Options: A) … (impact on time/cost/scope) B) … (impact on time/cost/scope) Ask: Please choose A or B by [date/time].

Escalate early (24–48h blocked). Late escalation is a reputation killer.

Definition of Done (avoid “almost ready”)

A task is Done only if:

implemented + reviewed + tested
accepted vs criteria (or stakeholder sign-off)
docs/notes updated (if needed)
deployed/handed over (if applicable)

Project close checklist

Confirm outcomes & metrics (what shipped / what changed)
Handover: owner, docs, runbook, support contacts
Retro: 3 wins, 3 lessons, 3 improvements
Archive RAID + final status + links

Spreadsheet columns (copy into Excel / Sheets)

For ICE/RICE/WSJF: Title · Problem/Stage · Reach · Impact · Confidence · Effort/Size · Score · Owner · Decision · Notes

For RAID: Type (R/A/I/D) · Description · Owner · Next action · Due date · Status · Link