---
name: "skill-content-case-study"
type: skill
description: 'Procedure for writing careerOS case studies and portfolio case cards from a truth capture — reading sources in order (truth captures, then case-study documents, then the published deck), choosing the output format, and applying the card rules the cOS specifications do not carry: zero or three proof points, a maturity score in points rather than percent, every count derived with its basis stated, and the read order deciding wherever sources disagree. Points at cos.output-specs, cos.case-study-task-standard, cos.case-study-qualification and cos.tov rather than restating them. Read when a case study or a portfolio Work card is being written or compressed to another format.'
license: CC BY-NC-SA 4.0
---

# skill-content-case-study

**Status:** Todo · proposed, not yet drafted · **medium candidate** (build only if the pattern recurs)
**Category:** Skill
**Proposed:** 2026-05-28

---

> *This document is a placeholder for a proposed skill. No draft exists yet. The content below captures why this skill was proposed and whether it warrants building.*

## What this skill would do

Encode the procedure for writing case studies in the cOS system — sourcing from a truth capture, picking the right format (snapshot / full / article / cv-entry per `cos.output-specs`), applying the writing standard, and validating against the quality check.

## When it would trigger

"Write a case study for X," "compress the JPMC case to a snapshot," "draft the cv-entry version of the website build" — any output generation that maps to the case study formats.

## Why it's a *medium* candidate (not high)

You already have `cos.case-study-task-standard`, `cos.case-study-qualification`, and `cos.output-specs` as detailed Linear docs. If those specs are thorough enough that following them is the procedure, a skill on top would be thin — it would say "for case studies, read these specs first, source from the tc, apply `cos.tov`" and not much more.

The reason to hold off until later: **premature skills encode the wrong thing.** Worth writing a couple of case studies (COS-186 is one queued) and seeing where things actually go wrong before deciding whether a skill earns its place above the specs.

## What it would reference

* `cos.output-specs` (format definitions)
* `cos.case-study-task-standard` (writing standard)
* `cos.case-study-qualification` (deciding *whether* something becomes a case study)
* `cos.tov` (tone)
* The truth capture being drawn from

## Open question

The honest open question: **does this skill actually do work above and beyond the specs?** Defer the build until that question can be answered with evidence — write a few case studies, see what's hard, then decide.

## Priority

Medium. Lower than `cos-linear-conventions`, `skill-content-truth-capture`, `website-build-delivery`. Medium was the holding position. The gate is met: `APP-3041` records the failure modes, so the card canon below is written rather than deferred. The placeholder framing earlier in this file is superseded by it.

## Portfolio case card

A card is a short entry on the Work index of aledpritchard.com. It is another output format off the same truth capture, which is why it sits here rather than in a skill of its own. The format set in `cos.output-specs` does not yet include it.

**Read sources in order.** cOS.Content truth captures (`tc.*`) first, then the case-study documents (`cs.*`), then the published deck. Fetch each document rather than recalling it.

**Where sources disagree, the read order decides and the basis is written.** This is the rule a card most needs. On the Dynamo card the maturity figure read 40 in the snapshot against 41 in the truth capture and the deck; team size read 2 in `tc.jpmc-team`, 5 counting designers only, and 1 on the CV; the role read "Design Lead" on the card against "VP Experience Design Lead" in the deck. Take the figure the read order gives, and state what it counts.

**Every count is derived and its basis is stated.** A team of five is five designers, not five people. A count without its basis is a count that gets corrected.

**State a maturity score in points, never percent.** A move from 41 to 62 is 21 points.

**The note is problem, then action.** Keep it light, because the case study carries the depth. One clear action, not a list of them.

**Lead with the headline metric, shown large.** Context and problem follow it.

**A card carries zero or three proof points.** Two is an incomplete set and four is a list.

**Context is not a result.** European expansion is the setting the work happened in, not a proof point. `cos.case-study-qualification` draws the same line for case studies: a traceable result, not a satisfied stakeholder.

**Show judgement, not polish.** A wrong turn, a cut, a trade-off with no clean option, what was left unfinished. `cos.case-study-task-standard` requires this of the narrative, deck and article; a card is not exempt for being short.

**Name who defined the problem, not only who solved it.**

**Where impact was not measured, state the metrics expected to move.** Do not reach for a number nobody took.

Keep the process at a high level and the card skimmable. An outside voice, a colleague quote or a testimonial, strengthens a card where one exists.

### What the specifications already carry

Read these rather than restating them.

* `cos.output-specs` — the format set, the shared spine, and the quality check
* `cos.case-study-task-standard` — the source hierarchy, the traceability rule, and the instruction to flag a gap rather than write around it
* `cos.case-study-qualification` — whether the work is a case at all, and which format it warrants
* `cos.tov` — register, and the words to avoid

The card rules above are the ones those four documents do not carry.
