AI Prioritization

AI Prioritization

Overview

This skill turns AI into a systematic prioritization engine — one that can evaluate projects against company goals, score them objectively, and surface which work deserves attention right now versus later. The underlying principle: prioritization is a continuous practice, not a one-time exercise. Design a system you revisit often, and let AI do the heavy lifting.

Core Rules

1. Anchor every prioritization decision to three axes: impact, urgency, and alignment. Work that scores low on all three is a candidate for deferral or elimination.
2. Urgency is not the same as priority. Just because something is due tomorrow does not make it the most important thing to do. Validate urgency by examining dependencies, not deadlines.
3. OKRs are only useful if you consult them. Load your company objectives into AI at the start of every prioritization session; never prioritize from memory alone.
4. Use a scoring model to remove subjectivity. Numbers make priorities legible and defensible to leadership. Let AI construct and run the model — you supply the criteria.
5. Always validate AI output against your own judgment. AI has no access to political context, interpersonal dynamics, or unspoken constraints you hold. It is a thought partner, not a decision-maker.
6. Communicate priorities with evidence, not assertions. "This is important" is not enough. Surface the scoring data, the OKR alignment, and the dependency chain when presenting to stakeholders.
7. Update AI as business context changes. Stale inputs produce stale rankings. Treat your AI tool as a team member who needs ongoing briefings, not a static calculator.
8. Close the feedback loop. After projects are completed, brief AI on what worked and what did not. This improves the quality of future prioritization cycles.

Workflows

1. Define the Prioritization Criteria

The problem: Teams often prioritize based on whoever is loudest or most recent, rather than on a consistent set of criteria applied equally across all work.

Before evaluating any project, define the three core dimensions you will use to assess priority:

• Impact — Does this drive revenue, grow users, or improve team effectiveness?
• Urgency — Does delaying this block another team or create downstream risk?
• Alignment — Does this ladder up to a stated company goal or OKR?

Prompt template:

These are our company goals for this period: [PASTE OKRs OR GOALS]
These are the projects currently under consideration: [LIST PROJECTS]
Using impact, urgency, and strategic alignment as evaluation criteria,
give me an initial ranking with a one-sentence justification for each project's position.

Rule: Run this prompt at the start of every planning cycle — quarter, month, or sprint. Never prioritize from memory.

2. Evaluate Urgency with AI

The problem: Everything is called urgent. Real urgency is defined by dependency chains, not by who sent the loudest message.

Use AI to distinguish genuine urgency from perceived urgency by examining what gets unblocked if you act now.

Prompt template:

A new project has been flagged as urgent: [PROJECT DESCRIPTION]
These are the other projects currently in flight: [LIST ACTIVE PROJECTS]
Identify any concrete dependencies this urgent item would unblock
if we prioritized it immediately. Are those dependencies significant enough
to warrant reprioritizing our current work?

Urgency validation checklist:

• Does acting on this immediately unblock another team or person?
• Does delaying this create a measurable risk to an active project?
• Is the deadline externally imposed (customer, contract) or internally invented?
• If we do this now, which current priority do we displace — and what is the cost of that?

Diagnostic prompt:

If we drop our current highest-priority work to address [URGENT ITEM],
what is the realistic downstream cost? List the dependencies we would delay
and the risk level of each.

3. Map Work to Strategic Alignment (OKRs)

The problem: OKRs are set at the start of the year, then ignored. When teams prioritize without reference to stated goals, effort diffuses across low-value work.

Use AI to run a ruthless alignment check: which projects on the list actually connect to a company objective, and which are orphaned work that has accumulated over time?

Prompt template:

These are our current company OKRs: [PASTE OKRs]
These are all the projects we are currently working on or considering:
[LIST ALL PROJECTS]
Classify each project as:
  - Directly supports an OKR (specify which one)
  - Indirectly supports an OKR (explain the connection)
  - No clear OKR alignment
Sort the output by alignment strength, strongest first.

Using the alignment map:

• Projects with direct OKR alignment → highest scheduling priority
• Projects with indirect alignment → second tier; review whether indirect connection is real
• Projects with no alignment → strong candidates for deferral or cancellation

Verification rule: Always review AI's alignment classifications manually. AI may misread the connection between a project and an OKR, especially for domain-specific work. If the classification looks wrong, correct it and feed the correction back to AI.

4. Quantify Impact Using a Scoring Model

The problem: Comparing projects by feel is unreliable and indefensible. A scoring model converts qualitative judgment into a structured number that can be compared, sorted, and explained.

Let AI build and run the scoring model — you define the factors.

Prompt template:

I want to score projects for prioritization using the following factors:
  - Strategic impact (1–5): How directly does this drive our top company goal?
  - Urgency (1–5): How much does delaying this harm active dependencies?
  - Effort-to-value ratio (1–5): How much output do we get relative to the work required?
  - OKR alignment (1–3): Direct / indirect / none

Here are the projects to score: [LIST PROJECTS WITH BRIEF DESCRIPTIONS]

Score each project on every factor, calculate a total, and rank them from highest to lowest.
Include a one-sentence rationale for any score you assign.

Score interpretation guide:

Rule: The scoring model is a decision aid, not a decision-maker. If your judgment conflicts with a score, investigate why. The discrepancy is usually useful signal.