Retention and proof
Write up a win you can prove.
Drafts a case study from a result your numbers log already holds: where things stood, what was done, and the outcome against its target, with every figure cited to its ledger row or another named source. The customer quote stays a marked placeholder until a real one arrives, and if no row matches, it stops and asks for the data instead of estimating.
- No logged row, no claim
- Quote left as a placeholder
- A judge tests each figure
At a glance
- Group
- Retention and proof
- Command
- /case-study
- Uses
- Your numbers log and work notes
- Writes
- A case study draft, a log row
- Sends
- Nothing. Drafts only.
01When to use it
Use it once a result is on record.
- A metric in your numbers log has moved a real distance from its baseline, and the win deserves a write-up.
- You want a proof piece for sales, your site or a pitch deck, made only from numbers you can defend.
- A client or project reached a milestone with an outcome someone measured.
Not for
- Drafting the request for the quote the story needs. Use /testimonial
- A scorecard of every metric for a month or quarter. Use /business-review
- A LinkedIn post about the result. Use /linkedin-writer
02What it needs
A logged result, and notes on the work.
- Your numbers log, with a row for the metric: date, baseline, current value, target, source and confidence. On an agency setup, the client goals file, for the active client only.
- Notes on the engagement, such as the project or content brief, or the client folder on an agency setup. Process documents and decision records help show what was actually done.
- Your voice files, or the client brand file on an agency setup, loaded before a word is drafted.
- The metric or row to feature, if you have one in mind.
- Your rules file. Anything it prohibits is turned down by name, and any condition it attaches is carried into the draft. If the file is missing, you are told, and no public claim is drafted until you reply.
03What it writes
A draft, its figures and a provenance row.
- final/case-study.md
- The draft, in three parts: the starting point, what was done, and the measured outcome from baseline to current against the target, each figure cited to its ledger row or source. A figure from an inferred or stale row carries a note to verify it before publishing, and a marked space waits for the quote.
- data/
- Any supporting figures.
- Operations/case-studies/{slug}.md
- Where the draft goes on an agency setup: anonymized by default, marked confidential with consent pending, and naming no other client.
- Ledger row
- A record that the result was packaged into a draft, added to the same log the numbers came from. Its note names the confidence of the row the numbers came from, and it never calls an outcome confirmed when that row does not.
04Limits
What it never does.
- Publish, post, send or email the case study.
- Invent, estimate or inflate a figure, or round past what the row states. No row, no claim.
- Write a customer quote. The placeholder stays until a real, attributed one is supplied.
- Say a result happened more often than the record shows. Logged once means told once.
- Fill a gap with something made up. A fact it has not read is named as missing.
- Change, reorder or delete a row that is already in the log.
What stays with you.
- Publishing it, after checking every flag it raised.
- On an agency setup, the client's consent before the draft is shown outside the agency.
- A real quote from a named person, to replace the placeholder.
- Deciding who signs off when your files disagree on the approver. The draft names both.
05Sample output
One draft, part by part.
From a made-up case study about a removals company. Figures show as labels here; a real draft prints each one exactly as its ledger row records it.
| Part | Draft line | Cites |
|---|---|---|
| Situation | Estimate requests sent on a weekend waited until Monday for an answer. | Project brief |
| What we did | A weekend reply rota, and an estimate form cut to five fields. | Decision, 2026-07-02 |
| Outcome | Weekend estimate requests answered within a day moved from [baseline] to [current], against a target of [target]. | Ledger, 2026-09-15 |
| Quote | "{TESTIMONIAL · pending from the operations lead}" | Placeholder |
- Situation Project brief
- Draft lineEstimate requests sent on a weekend waited until Monday for an answer.
- What we did Decision, 2026-07-02
- Draft lineA weekend reply rota, and an estimate form cut to five fields.
- Outcome Ledger, 2026-09-15
- Draft lineWeekend estimate requests answered within a day moved from [baseline] to [current], against a target of [target].
- Quote Placeholder
- Draft line"{TESTIMONIAL · pending from the operations lead}"
Flags in the draft
- NOTE: outcome confidence is inferred. Verify before publishing.
- The quote stays a placeholder until a named person supplies one.
- Status: draft. Publishing needs a person.
06Related skills
Often run beside it.
-
/testimonialDraft review requests and reuse replies Retention and proof -
/memoryRecord numbers, decisions and lessons as you go Core modules /fact-checkerCheck claims against their sources Build and utilities-
/landing-pageDraft a landing page with one action Content and publishing
07Questions
Before you run it.
What if the result is not logged yet?
It stops and asks you to set the baseline or record the result first. No figure is estimated to fill the gap, so the draft waits for the row.
Will it write the customer quote?
No. It leaves a clearly marked placeholder naming who the quote is pending from, and only a real testimonial, attributed to a person, takes its place.
Can an agency name its client?
Not by default. On an agency setup the draft describes the client by role, industry and size unless consent is on file, is marked confidential with consent pending, and is not treated as cleared for outside use until the client agrees.
Have us build it
We can build it for you.
Start with the free plan. A person names the first fix.



