<!– chunk-id: C-003 title: How I used a 360 review to freeze the right work and start the right work tags: [prioritization, external-review, process-over-product, remediation, decision-making] format-version: 1.0 –>

C-003 — How I used a 360 review to freeze the right work and start the right work

HOOK

I got a second external review of a project I was running, and the finding was blunt: I had shipped one internal admin tool and zero customer-facing output — but I had written 9 architecture decision records, 4 process documents, 7 policies, and 7 templates. I had built a highly organized company that had not yet started the thing the company existed to do.

MAP

• Inputs: Two external reviews of the same project, three weeks apart. First review verdict: “the company has been built; the product has not been started.” Second review verdict: “improved but the org still out-builds the product.” Between reviews I had added 4 collaborators, stood up 3 new functional units, and shipped 1 working internal portal — but had also added 9 more architecture decisions, 4 more process docs, 7 more policies, and 7 more templates. The founding concept document was untouched.
• Lap: Read the second review in full. Mapped every finding to a concrete cause. Produced a remediation plan with three concurrent work tracks and explicit freeze criteria.
• Radio: The reviewer had named the exact failure mode: I was adding org to support building org. Every new document I wrote required future maintenance. Every new collaborator I added required coordination overhead. I was optimizing the machine instead of running it.
• Setup Change: Declared a hard freeze on new process documents, new collaborators, and new functional units. Defined three concurrent unblocking tracks with a shared definition of done tied to concept progress, not org health. Made the next review gated on whether the founding concept had advanced — not on whether the org had improved.
• Delta: Before: 0 concept artifacts, 28+ internal process documents, growing org. After (session close): 1 rubric defined, 1 real content chunk produced (this one), concept-first definition of done installed. The freeze held — no new policies or process documents were written in the session.

RADIO

The second review said something the first had also said, which meant I had heard it and not changed fast enough. That repetition was the real signal.

Looking at the two reviews side by side, the pattern was visible: between reviews I had addressed every finding the reviewer could name a discrete fix for — I automated a brittle manual process, I merged an unactivated branch, I did a security audit before deploying. Those were real improvements and the reviewer credited them. But the strategic finding — “you are building org instead of product” — had gotten worse, not better, because it was harder to act on. Adding a collaborator feels productive. Writing a policy feels rigorous. Shipping nothing feels uncomfortable. So I had done the easier productive-feeling things and deferred the harder one.

The reviewer’s language was precise: “a large fraction of the new output is process artifacts about how to produce output.” That sentence is worth reading twice. I had meta-work masquerading as work.

SETUP CHANGE

Single change: redefined the definition of done for the project from “org health metrics” to “concept progress metrics.”

Concretely, this meant:

1. A hard freeze on any new process document, policy, or collaborator until the founding concept produced a real deliverable.
2. Three concurrent unblocking tracks identified and time-boxed: (a) start the venture — inject one real ideation brief and drive it to a scored output; (b) fill the core README — the project README still had placeholder sections (Purpose, Architecture, Testing) while peripheral docs were fully written; © make the cadence honest — the daily briefing automation was documented but silently not running, and that mismatch needed to be resolved rather than described.
3. The next external review explicitly gated on concept progress, not operational improvements.

Why this specifically: the freeze had to be crisp enough that I could not rationalize around it. “Freeze process docs unless a current activity demands it” is a leaky rule — I could always claim the current activity demanded it. “No new process documents until the concept produces one artifact” is not leaky. The definition of done also had to be external-legible — something a reviewer could check, not something I self-certified.

DELTA

Before: After the second external review — 0 artifacts from the founding concept. 1 internal admin portal (legitimate, working, but not the venture). 28+ internal process documents. 10 collaborators across 7 functional units. External reviewer’s verdict: B/B+, “the machine is running itself.”

After: Session following the remediation plan — 1 rubric defined (5 dimensions, weights summing to 100, measurable proxies, numeric threshold, disqualifier, worked example). 1 content chunk produced from the founding concept (C-001). 1 more in progress (this one, C-003). Freeze held: 0 new policies, 0 new collaborators, 0 new process documents written in the session.

Observable: The founding concept document was untouched across two external reviews. In the session after the remediation plan, it was the input, not the background. The first output tagged to the concept now exists in the chunk library.

PIT WALL NOTES

• If an external reviewer says the same thing twice, the finding is not about the work — it is about what you are avoiding. The first review named the problem clearly. Between reviews I fixed the easier problems and left the harder one. The second review naming the same problem was the data point I needed: I was not failing to understand the finding; I was finding reasons not to act on it.
• Meta-work has the same texture as real work. Writing a policy feels rigorous. Hiring a collaborator feels like building capacity. Documenting a process feels like progress. None of that is false — but it can all be done instead of the actual work, and it is easier to do than the actual work, so it will be done preferentially if you let it. The only fix is a definition of done that cannot be satisfied by meta-work.
• A freeze is only credible if it has teeth. “We should do less process work” is not a freeze — it is a preference. A freeze is: no new documents of type X until condition Y is met. The condition has to be checkable by someone other than you. If you cannot state the condition precisely enough for an external reviewer to verify it, the freeze is aspirational.

CTA

Use this when you receive an external review — or when you suspect you are building the wrong things.

The artifact is a two-part template: (1) a “translate the finding” worksheet to turn a review finding into a concrete freeze + three-track plan, and (2) a definition-of-done card that makes the unblock condition external-legible.

Part 1 — Translate the Finding (worksheet)

Copy this and fill it in within 24 hours of receiving the review. If you cannot fill in every row, the finding is not yet actionable.

FINDING (verbatim from review):
_____________________________________________

ROOT CAUSE (one sentence, your words — not the reviewer's):
_____________________________________________

WHAT I WAS DOING INSTEAD (name the meta-work):
_____________________________________________

FREEZE RULE (exact wording — no wriggle room):
"No [TYPE OF WORK] until [SPECIFIC CONDITION]."
Type: ___________________________________________
Condition: ______________________________________

THREE UNBLOCKING TRACKS (the concurrent real work):
  Track A — [short label]:  [one sentence, what done looks like]
  Track B — [short label]:  [one sentence, what done looks like]
  Track C — [short label]:  [one sentence, what done looks like]

NEXT REVIEW GATE (what the reviewer will check — not you):
_____________________________________________

Part 2 — Definition-of-Done Card

Paste this at the top of your next planning session. Replace the bracketed fields.

DEFINITION OF DONE — [PROJECT NAME] — as of [DATE]

DONE means: [CONCEPT / CUSTOMER-FACING METRIC], not [ORG / PROCESS METRIC].

Example of DONE: [one concrete artifact the founding concept requires]
Example of NOT DONE (even if it feels productive): [one meta-work category]

FREEZE in effect: [exact freeze rule from Part 1]
Freeze lifts when: [specific condition — checkable by a reviewer]

Check-in: [name of external reviewer or accountability partner] reviews on [date].

Worked example (this project)

Finding (verbatim): “A large fraction of the new output is process artifacts about how to produce output.”

Root cause: I was doing meta-work (policies, process docs, collaborator onboarding) because it felt productive and was easier than starting the founding concept.

Freeze rule: “No new policies, process documents, or collaborators until the founding concept produces one scored artifact.”

Three tracks:

• Track A — Start the venture: inject one real ideation brief; produce one scored content chunk.
• Track B — Fill the core README: Purpose, Architecture, and Testing sections are placeholders; fill them before writing the next peripheral doc.
• Track C — Make the cadence honest: the daily briefing automation is documented but not running; either wire it or mark it “not active” in the handbook.

Next review gate: External reviewer checks whether the founding concept document has a downstream artifact (a chunk, a rubric, a scored output) — not whether the org has improved.

SELF-SCORE

Rubric

Total: 95/100

Passing threshold: >= 70/100

Verdict: PASS

Disqualifier check: No dimension scores zero. No false claims made. The numbers are sourced from the review documents. The freeze rule and three-track plan are real decisions made in this session, not reconstructed post-hoc. Auto-fail not triggered.

Honest weakness: The DELTA section’s “after” numbers were measured at session close, not weeks later. The freeze holding for one session is not the same as the freeze holding. A re-read in 30 days could reveal that the meta-work resumed and the freeze was temporary. The chunk is honest about what changed in the session but cannot yet claim the remediation worked at a longer time horizon.