Key takeaways
- Agree the criteria and their weights before anyone sees a single project score. Criteria set after the scores are visible get reverse engineered to protect favorites.
- Invite people who can commit money and people. A room full of delegates produces a recommendation, not a decision, and recommendations get relitigated.
- Name the quorum and the veto voter in writing beforehand. Without both, one absent executive can reopen the whole ranking a week later.
- The ranking is not the output. The capacity line drawn across it is, because that line is the only place where the organization actually says no to something.
- Track decision reversal rate: the share of workshop decisions overturned outside the room within a quarter. Above 20 percent means the wrong people were in the room.
A project prioritization workshop is a scheduled working session where the people who control funding and staffing rank a defined list of candidate projects against criteria they agreed in advance, then draw a line at the point where capacity runs out. Everything above the line gets resourced. Everything below it gets an honest no, or an honest not this quarter.
That is the last definitional paragraph on this page. The rest is the mechanics of running one: what you send out beforehand, how the three hours are spent, who has to be there, what happens when two projects tie, and what happens when the CFO wants their project moved up regardless of its score.
The pre-work that decides whether the workshop works
The workshop is won or lost in the week before it. A session that opens with people reading project descriptions for the first time will spend its entire budget on comprehension and none on judgment. Send a data pack at least five working days ahead, and make it short enough that people actually read it.
One page per candidate project, no exceptions, containing only what changes a ranking decision:
| Field | What it holds | Why the ranking needs it |
|---|---|---|
| Outcome in one sentence | The measurable change the business gets, not the deliverable. | "Replace the CRM" cannot be compared to anything. "Cut quote turnaround from 6 days to 1" can. |
| Total cost to done | Internal effort plus external spend, across the whole life of the work. | Ranking on first-year budget systematically favors projects that hide their tail. |
| Peak specialist demand | The scarcest role it needs, and in which months. | Two projects can each look affordable and still be undeliverable together. |
| Hard date, if any | Regulatory or contractual deadline, with the source named. | Separates genuine must-do work from urgency someone asserted. |
| Dependencies | What must land first, and what breaks if this slips. | A high scoring project that depends on a low scoring one is not actually high scoring. |
| Sponsor | A named executive who will own the benefit. | Work without a sponsor has nobody to defend it later or accept the number. |
Swipe to see more →
Two rules keep the pack honest. First, whoever assembles it fills every field or marks it unknown, and unknown is allowed. A blank invites the room to assume the best case. Second, the pack is frozen. Late additions do not get scored, because a project that arrives the night before has not been through the same scrutiny as the rest and will win on novelty.
How to set the decision criteria before the session
Criteria have to be agreed and weighted before any project is scored against them. This is the single highest leverage rule in the whole process. Once people can see how their project scores, criteria stop being a measuring instrument and become an argument, and the debate quietly moves from "is this the right project" to "is this the right weighting", which is unwinnable.
Four to six criteria is the working range. Fewer and everything scores the same. More and the weights get so thin that the bottom two criteria cannot change any outcome, which means you spent time scoring them for nothing.
Run criteria setting as its own short session two weeks earlier, with the same people. Open it by generating far more candidate criteria than you need, ideally in silence and independently so the most senior voice does not anchor the room, then cluster and cut. If you want to widen that first pass beyond what six people can produce in ten minutes, it helps to push one prompt through a structured divergent round and use the output as raw material to cluster, not as the answer. Then force the weights to sum to 100 and make people spend that budget. A criterion nobody will spend points on is a criterion nobody actually believes.
For how to write criteria that discriminate rather than flatter, and the common ones worth stealing, see project prioritization criteria.
A project prioritization workshop agenda
Three hours, timeboxed, with the scoring done live. The times below are for a list of 15 to 25 candidates, which is the range where this format works. Above 30 candidates, screen first and bring the survivors.
| Time | Segment | Output |
|---|---|---|
| 0:00 to 0:10 | Restate the criteria, weights, quorum and veto voter. No debate, just confirmation. | Everyone bound to the same rules before scoring starts. |
| 0:10 to 0:25 | State the capacity envelope: money and the two or three scarcest roles, in numbers. | The room knows roughly how many projects can survive. |
| 0:25 to 1:10 | Independent silent scoring against each criterion. | A raw score per project, uncontaminated by the room. |
| 1:10 to 1:25 | Break while scores are collated and the spread is calculated. | Ranked list plus a disagreement flag on each project. |
| 1:25 to 2:10 | Discuss only the projects where scores disagreed most. Rescore those. | Converged ranking, with the reasoning captured. |
| 2:10 to 2:35 | Draw the capacity line. Walk down from the top until money or the scarce role runs out. | The funded set and the explicit not-now set. |
| 2:35 to 2:55 | Test the line against dependencies and hard dates. Adjust and record why. | A ranking that survives contact with sequencing. |
| 2:55 to 3:00 | Confirm the decision record and who communicates what by when. | Signed decisions, named owners, dates. |
Swipe to see more →
The segment people cut when they run late is the silent scoring, because it feels like dead air. Cutting it is what turns the workshop into a discussion that the most confident speaker wins. Cut the discussion block instead and carry the two most contested projects to a follow up.
Who should be in the room
Invite the smallest group that can commit both money and people, plus the person who has to deliver the answer. In practice that is five to nine decision makers. Below five the decision looks like one person's opinion. Above nine, silent scoring still works but the discussion block stops converging.
| Role | Why they are there | What goes wrong without them |
|---|---|---|
| Budget holders | They can actually fund the top of the list. | The ranking becomes a wish, and funding gets decided somewhere else later. |
| Resource owners for scarce roles | They know whether the capacity envelope is real. | The line gets drawn on money alone and the plan fails on people. |
| Project sponsors | They defend their case and accept the outcome in person. | Absent sponsors reopen the decision by email. |
| PMO facilitator | Runs the process, holds the timebox, does not score. | Whoever runs it also wins it, and everyone can see that. |
| A named veto voter | Usually the most senior person. Breaks deadlock fast. | Deadlocks get parked, and parked deadlocks default to the status quo. |
Swipe to see more →
Set the quorum in writing before invitations go out, and say plainly what happens if it is not met: the session is moved, not run with substitutes. Delegates are the most common cause of a decision being reversed afterwards, because a delegate can agree to a ranking but cannot bind the person who sent them.
The capacity line is where prioritization actually happens
A ranked list is not a prioritization. Every organization can produce a ranked list and then start all of it anyway, which is the most common failure mode in portfolio management and the reason prioritization has a reputation for being theater.
Prioritization happens at one moment: when someone draws a line across the ranked list at the point where money or the scarcest role runs out, and the projects below it are told no. Walk down the list in rank order, subtracting cost and specialist months as you go, and stop at the first project that does not fit. Do not skip past it to a cheaper project further down. Skipping is how the line stops meaning anything.
Test the line against two things before it is final. Dependencies: if a funded project depends on an unfunded one, either pull the dependency above the line or move the dependent below it. Hard dates: a regulatory project with a real deadline sits above the line regardless of score, and the room should say out loud that it is being exempted rather than quietly inflating its score to justify it. Draw the envelope from real availability rather than headcount, which is what resource capacity planning exists to produce.
How do you break a tie in a prioritization workshop?
Break ties with a pre-agreed tiebreak order, applied in sequence, not with more discussion. Discussion on a tie takes fifteen minutes and resolves in favor of whoever cares most, which is not a portfolio criterion. Write the order into the terms of reference before the session so it is never negotiated in the moment.
A tiebreak order that holds up in practice: shortest time to first measurable benefit, then lowest demand on the scarcest role, then the one that unblocks the most other work, then the veto voter decides. The first rule breaks most ties on its own, and it biases the portfolio toward earlier feedback, which is usually the right bias when two options genuinely score the same.
What if a sponsor overrides the ranking?
Let the override happen, record it as an override, and name what it displaced. Executives are allowed to make decisions the model did not, and a process that forbids it will simply be bypassed. What you cannot allow is an override disguised as a score, because that corrupts the instrument for every future session.
The protocol is one line in the decision record: project X moved above the line by name, decided by name, displacing project Y. That is it. No argument in the room. The cost of an override is transparency about its consequence, and in most cases writing the displaced project's name down is enough to make the override voluntary. When it is not, the portfolio still got a decision, which is better than a stalemate, and the record is what your portfolio governance forum reviews next cycle.
The decision record the workshop has to produce
If the only artifact is a ranked spreadsheet, the workshop will be relitigated. The record needs the reasoning, because the reasoning is what stops the same debate reopening in six weeks with the same people.
- The criteria and weights used, as a snapshot, since they may change next cycle.
- The final ranked list with each project's score.
- Where the line was drawn, and the constraint that drew it (money, or which specific role).
- Every project that moved for a reason other than its score, with the reason and the decider named.
- The not-now list, with the review date when each gets reconsidered.
- Who communicates the outcome to each affected team, and by when.
The not-now list matters more than teams expect. A project told no with a date attached behaves very differently from a project told no with silence. The second one gets restarted quietly under a different name, and you meet it again next cycle as a new candidate.
Decision reversal rate: the number that tells you the workshop is real
Most PMOs measure prioritization by whether the workshop happened. A more useful measure is what share of its decisions survived. Decision reversal rate is the percentage of workshop decisions that were overturned, materially altered, or quietly ignored outside the room within one quarter of the session.
Count a reversal when a below-the-line project starts anyway, when an above-the-line project is defunded without a subsequent governance decision, or when the agreed sequence is abandoned. Divide by total decisions made in the session.
| Reversal rate | What it usually means | What to change |
|---|---|---|
| Under 10 percent | The right people were in the room and the decisions stuck. | Nothing. Keep the attendee list stable. |
| 10 to 20 percent | Normal. Some reversals are legitimate responses to new information. | Check the reversals were governed, not informal. |
| Above 20 percent | The room lacked authority, or the capacity envelope was fiction. | Fix the invitation list first, the numbers second. |
| Above 40 percent | The workshop is theater and people have noticed. | Stop running it until a budget holder will own the outcome. |
Swipe to see more →
These bands are operating rules to calibrate against your own history, not measured industry constants. Track your own number for three cycles before you trust the threshold you set.
How long should a project prioritization workshop be?
Three hours for 15 to 25 candidate projects, run once per planning cycle, with a separate 60 to 90 minute criteria session two weeks earlier. Splitting criteria from scoring is what protects the criteria from being tuned to the results, and it is the difference between a repeatable process and a single long meeting that exhausts everyone.
Longer sessions do not produce better rankings. Past about three hours, scoring quality drops and the room starts optimizing for finishing. If the list is genuinely too long, run a cheap screening pass first using two criteria and a simple grid, and bring only the survivors to the full session. The impact effort matrix is the usual screen for that, because it sorts a long list fast without pretending to be precise.
Five ways prioritization workshops fail
Every one of these is recoverable, and every one of them is easier to prevent in the invitation than to fix in the room.
- Criteria agreed after the scores are visible. The weights get tuned until the expected answer appears, and everyone in the room knows it happened.
- No capacity envelope. Without a number to run out of, the line never gets drawn and the session produces a ranked list that funds everything.
- Delegates instead of decision makers. Produces a recommendation. Recommendations have no authority and get reopened by the people who did not attend.
- Scoring out loud. The first number spoken anchors every number after it. Silent independent scoring costs 45 minutes and removes the effect entirely.
- No not-now list with dates. Rejected work goes underground, consumes capacity informally, and reappears next cycle with a new name and a new sponsor.
Frequently asked questions
What is the difference between a prioritization workshop and a portfolio review meeting?
A prioritization workshop decides what to start, ranking candidate projects and drawing a funding line, and it runs once per planning cycle. A portfolio review meeting inspects work already underway, checking progress, risk and continued value, and it runs monthly or quarterly. One selects the portfolio, the other steers it.
Who should facilitate a project prioritization workshop?
Someone from the PMO who does not score and has no project in the list. The facilitator owns the timebox, enforces the criteria, and captures the decision record. If the person facilitating also sponsors a candidate project, the room will discount every process call they make, whether or not the call was fair.
How many projects should you score in one workshop?
Fifteen to twenty five in a three hour session. Below fifteen the format is heavier than the decision needs and a shorter meeting will do. Above thirty, scoring quality collapses in the second hour, so screen the list down first with a fast two-criterion pass and bring only the serious contenders.
Should projects already in flight be included in the workshop?
Yes, at least once a year. A portfolio that only ever ranks new candidates accumulates in-flight work that nobody has re-justified since it started, and that work is consuming the capacity the new candidates are competing for. Ranking active projects alongside new ones is uncomfortable and it is the only way stopping ever becomes a real option.
What scoring scale should the workshop use?
A 1 to 5 scale per criterion, with each level written down as an observable claim rather than an adjective. Wider scales suggest a precision the inputs do not have. What matters far more than the range is that every scorer reads the same definition of a 4, which is the work that building a scoring model is really about.
Can you run a project prioritization workshop remotely?
Yes, and silent scoring is easier remotely than in person because everyone submits privately by default. The two things that need extra attention are the timebox, which slips more on video, and the discussion block, where you should call on people by name in a fixed order so the quietest resource owner is heard before the ranking converges.
How often should a prioritization workshop run?
Once per funding cycle for the full session, which is quarterly in most organizations and annually in those with fixed budgets. Between full sessions, handle new arrivals through the standing intake route and score them against the same criteria, so a project that appears in month two is measured the same way as one that appeared in month one.
Where the workshop sits in the wider process
The workshop is one step in a chain, and it fails quietly when the steps around it are missing. Candidates should reach it through a consistent front door rather than by executive whisper, which is what a project intake process provides, and the form behind it standardizes what each candidate has to declare. The models the room scores against come from project prioritization frameworks, and the grid and formulas live in the project prioritization template.
Downstream, the funded list has to become a resourced plan, which means reconciling it against real availability and named specialists rather than headcount totals. From there the decisions are steered, not repeated: the ranked portfolio and its capacity line feed how to prioritize a project portfolio across cycles, and the whole discipline sits inside project portfolio management.
Last updated August 2026.