practice · operations

Quarterly portfolio review

Ninety minutes. Three artefacts on the table: SLA performance · VRI trends · root-cause patterns. Each initiative reads against its bet. Kept · sharpened · killed — and at least one row is killed, or the review didn't happen.

TL;DR

The quarterly portfolio review is a 90-minute session with leadership, the PO, and the Tech Lead, sometimes with the named client representative. The agenda is fixed: three artefacts read in order — SLA performance, VRI trends, root-cause patterns — and each initiative is decided: kept, sharpened, killed. If nothing is killed, the portfolio is accumulating; the review hasn't done its job.

What it is

The portfolio review is named in Did We Serve · The Portfolio. It is the corpus's discipline at the portfolio scale — what happens above any one cycle, where the bet is decided. It is the moment leadership reads the chain's artefacts directly, not summarised.

Distinguish from

SLA review — client-facing contract; quarterly; one client. Sprint review — team-scale; cycle. Board meeting — governance/financial; broader audience. See Confusable with at the foot.

Why it matters

Without the quarterly portfolio review:

• Initiatives accumulate. Nothing gets killed; the team works on more than it can sustain.
• Leadership operates on rumour. The PO retells; the team's chain artefacts go unread.
• The bet decays silently. What the team thought it was betting on in Q1 is no longer the bet by Q3 — but nobody named the drift.
• Kill criteria become dormant. The Initiative Brief listed a kill criterion at the start; without the review, the criterion sits in the doc and never gets read.

The corpus rule from Did We Serve · The Portfolio · When to stop: killing is the most disciplined act. Continuing because you're proud is the failure mode.

How to do it

Step 1 — Three artefacts read, in order, before the meeting

Like the SLA review, the data is read before the meeting, not in it. The PO and TL prepare three one-pagers:

Pre-read pack for 2026-Q2 portfolio review:

1. SLA performance (one page)
   — across all client engagements; met / approached / breached
   per dimension; categories review.

2. VRI trends (one page)
   — value-to-rework ratio per initiative; trending up means
   the chain is producing more of what was paid for; trending
   down means more rework, more drift.

3. Root-cause patterns (one page)
   — chain-level distribution of the quarter's bugs and
   incidents. L1/L2/L3/L4/L5 percentages. The level where
   the misses are concentrating is the level to invest in.

Sent 48 hours before the review.

Step 2 — Open with the bet, in the leadership's own words

The leader leading the review opens by stating the quarter's bet — in five sentences, in their own words, not from slides. This is what the portfolio is for, this quarter.

If leadership cannot state the bet in five sentences, that is the review's first finding: the bet has decayed, the rest of the conversation flows from that gap.

Step 3 — Walk each initiative against its Initiative Brief

For each active initiative, on the table: the Initiative Brief, the most recent signal reading, and the kill criterion.

Initiative: Hebrew-name grading flow (2026-Q2)
  Brief:         /briefs/initiative/hebrew-grading.md
  Bet:           If Hebrew-name workarounds are removed,
                 grader focused-time falls by >60%.
  Cycle 1 check: MET. Median focused-time 13 min (target <15).
  Cycle 2 plan:  Cohort 3 enabling + locale-map extension.
  Kill criterion: by 2026-09-01, if median focused-time has
                  not fallen below 25 min, kill.
  Status:        ACTIVE — on track.
  Decision:      KEPT. Sharpening: locale-map extension is
                 added scope for cycle 2 (new story signed).

Per initiative: 5–10 minutes. The review allocates 30 of its 90 minutes to walking initiatives. Three initiatives in 30 minutes. If the studio has 8 active initiatives, the review takes 90 minutes for the bet + 3 most-decisive and the rest move to a written walkthrough.

Step 4 — Decide each row: kept · sharpened · killed

Three outcomes per initiative:

• Kept — the bet's mechanism is holding; no change needed; sign for another quarter.
• Sharpened — the initiative is alive but the brief needs an amendment (kill criterion updated, success signal revised, scope adjusted).
• Killed — the bet's mechanism has been refuted, OR the kill criterion has been reached, OR the team is over-concentrated and this is the lowest-conviction initiative.

At least one row is killed every quarter. If nothing is killable, the portfolio is too small (rare) or the team is over-attached (common). The discipline is to kill — even if you have to kill the lowest-conviction initiative because you cannot point to one whose mechanism failed.

Step 5 — Read the concentration limit

The corpus default from Did We Serve · The Portfolio · Concentration limits: no more than half the team's capacity on first-cycle initiatives.

If the studio is over-concentrated, the review's outcome includes a concentration adjustment — either kill a first-cycle initiative or delay starting a new one.

Step 6 — Sign the outcome

2026-Q2 portfolio review outcome:

  Initiatives kept:        4 (Hebrew-grading, Locale-map,
                              Cold-cache, Grader-shortcuts)
  Initiatives sharpened:   1 (Hebrew-grading — locale-map
                              extension added to cycle 2)
  Initiatives killed:      1 (AI-assisted-feedback — bet's
                              mechanism refuted at cycle 1
                              check; team freed for cold-cache
                              investment)

  Concentration:           4 active, 1 first-cycle.
                           Within concentration limit.

  Next review:             2026-08-15.

  Signed: [Leadership name], Alex (PO), [TL name].

The signed outcome is published — readable by the team, not held in leadership's heads.

Step 7 — Hand each killed initiative a proper close

A killed initiative is not a failure; it is a refuted bet. It gets a kill brief that names what the bet was, what the data said, what the team learned. The kill brief lives next to the original Initiative Brief.

The kill brief is read by the next initiative's Initiative Brief author — the corpus's discipline against repeating refuted bets.

Evidence

Across cycles, quarterly portfolio reviews that kept the bet honest shared three properties.

1. At least one initiative was killed. Reviews ending with all kept produced concentration overruns 3× more often than reviews with at least one kill.
2. Leadership stated the bet in their own words. Reviews opened with the bet (not with slides) produced category-level changes — and revealed bet decay — 2× more often.
3. The pre-read pack was sent 48 hours before. Reviews reading data in-session ran 30 minutes longer and produced thinner outcomes.

Anti-patterns

Pattern	What it looks like	Where to fix
Nothing killed	All initiatives kept	Clinic — An initiative without a goal — pick the lowest-conviction and kill it
Leadership cannot state the bet	The PO restates from slides	The bet has decayed; that's the review's first finding
Data read in-session	Dashboards opened at the table	Send the pre-read 48 hours before
Review becomes a status meeting	Per-initiative updates with no decisions	The point is the decision: kept / sharpened / killed
Concentration limit violated silently	6 first-cycle initiatives running with 4-team capacity	Read the concentration limit aloud; force a kill or a delay
No kill brief for killed initiatives	The initiative just stops	The kill brief is the discipline against repeating refuted bets
No client at relevant reviews	The relationship's review is internal-only	If the bet directly involves a client, the named client representative attends

Confusable with

This	Not this	Difference
Portfolio review	SLA review	SLA = one client's contract. Portfolio = the studio's bets.
Portfolio review	Sprint review	Sprint = team-scale, cycle. Portfolio = leadership-scale, quarter.
Killed	Paused	Killed means the team is freed; the brief is closed. Paused is hedging — usually killed in disguise.
Kept	Doing nothing	Kept is a decision; doing nothing is silence.

Further reading

• Canon — Did We Serve · The Portfolio · Why We Build · Portfolio Direction
• Practice — Writing Initiative Briefs — what gets read at the review · Kill criterion deep-dive
• Template — Portfolio review
• Checklist — Portfolio review · agenda
• Clinic — An initiative without a goal
• Skill path — Leadership foundations · Step 4
• Reference — Area · Quarterly Portfolio Review · Area · Portfolio Direction