This guide gives you a reusable starting pattern. It is designed to help you see the work more clearly; it is not a substitute for judgment, source checking, or responsibility for the result.
Prepare the inputs before you ask for output.
The model only sees what you give it. Spend a few minutes naming the reader, desired result, and uncertain information. This makes a first draft easier to assess and reduces the need for decorative rewriting later.
Make the evidence trail reviewable.
Translate a reviewed evidence record into a smallest credible move while making the conditions that could change it visible.
Inputs required
- Reviewed evidence record. A brief or matrix with source labels, dates, and limitations—not an untraceable summary.
- Decision and authority. The choice supported and the role that may approve, defer, or redirect it.
- Operational constraints. Known capacity, dependencies, access, approvals, and boundaries for a reversible first move.
Method sequence
- Bound. State the decision and distinguish proposed work from approved work.
- Select. Choose one move that reduces a decision-critical uncertainty or verifies a dependency.
- Challenge. Record the evidence gap or observation that would change the priority.
- Gate. Define who reviews which evidence at the checkpoint and what decision follows.
Public sources used in this guide
- U.S. Government Accountability Office (GAO): Evidence-Based Policymaking: Practices to Help Manage and Assess the Results of Federal EffortsPublic-source walkthrough: planning, evidence-building, evidence use, and continuous learning.
Illustrative composite — not a client case
A verification step can be a priority without pretending to be a product commitment
This fictional record models a reversible first move and deliberately contains no user metrics, product result, or claimed outcome.
- Limited evidence
- Two labeled support notes report the same friction; technical cause and frequency are unknown.
- Smallest move
- Reproduce the issue against an agreed sample with explicitly available technical review.
- Disconfirming condition
- The issue cannot be reproduced using the agreed method, or required access is not available.
- Review gate
- The decision owner reviews the reproduction record before approving, deferring, or redefining broader work.
This is a fictional composite for method illustration. It is not a customer report, a test result, or a promised delivery plan.
Original working artifact
Reversible-priority card
Copy this original decision record alongside the downloadable priority-plan workbook; it makes the review gate explicit before work expands.
# Reversible-priority card ## Decision and status - Decision this supports: - Decision owner: - Approval status: proposed / approved / unknown ## Evidence carried forward | Claim | Source label and date | Limitation | Why it supports only this next move | | --- | --- | --- | --- | | | | | | ## One bounded priority - Smallest next move: - Scope boundary: - Dependency and owner status: - Disconfirming condition: - Work to defer: ## Review gate - Evidence to review: - Who reviews it: - Decision after review: continue / pause / redefine - Date only if explicitly agreed:
Give the task a useful brief.
Turn the reviewed evidence record below into a priority plan. Preserve uncertainty and do not treat a proposed plan as an approved commitment. Decision to support: [one decision] Decision owner and approval status: [known / unknown] Evidence that is sufficient for a next check: [claim | source label | date | limitation] Evidence gaps that could change the choice: [paste] Options considered and trade-offs: [paste] Constraints and dependencies: [capacity, approvals, access, timing] Return: (1) decision and current approval status, (2) one smallest reversible priority, (3) evidence supporting that next move with labels, (4) disconfirming condition or evidence gap, (5) dependency and explicit owner status, (6) review checkpoint and what it decides, (7) work to defer, and (8) a confirmation question for the decision owner. Do not invent priority, urgency, capacity, owners, dates, approval, evidence, or results.
Replace every bracketed field with your real context. Read the output before reuse.
Four steps that keep the result usable.
- 1
Start with the decision and current approval status, not a backlog.
- 2
Carry forward the evidence labels, source dates, and limitations that make the next move defensible.
- 3
Choose a first move that can be reviewed, paused, or reversed before wider work starts.
- 4
Name the disconfirming condition: the observation or missing evidence that would change the priority.
- 5
Record the dependency, checkpoint, and decision owner as fields—not optimistic assumptions.
A research brief is an input, not a commitment
A brief explains what is known, limited, and worth deciding. A priority plan adds a different contract: what is the smallest move justified now, what would change that choice, and who may approve broader work? If the plan begins with tasks, it can hide the fact that the decision has not been made. Keep the decision and approval status at the top of the record so a reader can see whether the plan is exploratory, proposed, or authorized.
Carry evidence forward as a planning constraint
Evidence does not become stronger because it appears in a plan. Preserve the source label, date, and limitation beside each claim that supports the next move. A reported issue can justify a scoped verification; it does not establish its cause, prevalence, or priority over every other problem. If the limitation could alter the move, it belongs in the plan’s disconfirming-condition field, not in a footnote.
Choose a reversible move instead of an impressive roadmap
A reversible move has a defined scope, a low-cost stopping point, and a result that informs the decision. Examples include reproducing an issue with a defined sample, checking an access dependency, or validating a required approval. It is not automatically a pilot, an experiment, or a release. The useful test is whether a reviewer can name what will be learned, what will not be decided yet, and how the work can stop if the premise fails.
Name the condition that would change the priority
Every credible plan should include a disconfirming condition. It might be “access to the required record is not granted,” “the reported issue cannot be reproduced using the agreed method,” or “the decision owner does not accept the trade-off.” The condition is not pessimism. It protects the team from continuing because a draft made a path feel inevitable. If no condition could change the plan, the task may be too vague or may already be a commitment that needs explicit approval.
Public-source walkthrough: plan, build evidence, use it, then learn
The U.S. Government Accountability Office’s public guide groups 13 evidence and performance-management practices across planning for results, assessing and building evidence, using evidence, and fostering continuous learning. That is a useful method boundary for a small planning record: write the intended result, state what evidence exists, use only enough evidence for the next bounded move, and include a review point. The guide is not an endorsement of any private-team priority or a prediction that a proposed action will succeed.
Illustrative composite: verify before promising a fix
Consider a fictional record in which two labeled support notes mention slow exports, but no engineer has reviewed the relevant path and capacity has not been approved. A plan can propose one bounded check: reproduce the issue against a defined sample with an engineer who is explicitly available. The plan carries the two notes as limited evidence, names access to the record as a dependency, and sets a checkpoint to decide whether a scoped fix is warranted. It must not say “ship export performance work this week.” This is an illustrative composite, not a customer case or a reported test result.
Use AI to test the plan’s missing fields
Ask an AI system to identify missing owner status, unsupported urgency, tasks that imply approval, and sentences that lack a disconfirming condition. Do not ask it to choose which trade-off the organization should accept. The model can make plan ingredients inspectable; a person with authority must decide the trade-off, capacity, and scope.
End at a review gate, not an unearned due date
A checkpoint is not a vague “follow up later.” It names the evidence to review, the decision it will inform, the role that will review it, and the condition under which the work pauses. If a date was not agreed, do not manufacture one. The plan is useful precisely because it gives a decision owner enough structure to confirm, defer, or revise the next move without confusing a planning draft with a promise.
Run a human check.
- Does the plan state the decision and approval status rather than merely list activity?
- Can a reader trace each decision-critical planning claim back to a labeled evidence record?
- Is the first move small enough to pause or reverse before wider work starts?
- Is a disconfirming condition named and genuinely capable of changing the priority?
- Are dependencies, owner status, and review gate visible rather than assumed?
- Has the accountable person confirmed the trade-off before the plan becomes a commitment?
Keep your human check visible.
Mark the checks you have personally reviewed. This is a local reading aid—not evidence that a claim, commitment, or decision is correct.
This checklist is stored only in this browser when storage is available. It is not sent to Knexio or used for analytics.
Download, preview, or print the research-to-priority plan.
Choose an editable Excel workbook, a CSV priority table, or a plain Markdown record. All formats keep source-aware evidence separate from the decision, one modest priority, dependencies, review condition, and human check. Downloads stay in your browser and do not require an account.
Excel includes: Start here, Priority plan, and Evidence notes worksheets. The CSV and print preview are simple priority-table records.AI is strongest here when it makes missing information, structure, and options easier to see. The moment an output becomes a claim, commitment, or decision, bring a person back into the loop.
What this page specifically establishes
This page connects reviewed evidence to a reversible priority plan while keeping uncertainty visible; it does not claim that a proposed priority will produce a business outcome.