Reusable Source-Derived Voice Profile

These examples and illustrative results are independently authored teaching materials, not measured model results.

Use case

An author supplies five short samples, and you need release notes and a newsletter for one launch. The shortened teaching samples are S1 ‘Bulk export added. Administrators can enable it in settings.’; S2 ‘Checks cover three input classes; mobile is untested.’; S3 ‘Read the rules, then choose the next step.’; S4 ‘Only ordering changes; the interface stays the same.’; S5 ‘Two of ten checks failed; records are in the appendix.’

Mechanism

Extract evidenced rhythm, claim strength, transitions and vocabulary, citing a sample for each trait. Do not turn an absent behavior into a permanent prohibition. Separate public copy from private working messages when the set contains both. Condense confirmed traits into a short profile reused by both drafts; medium differences change necessary context only.

Bad example

Use samples S1–S5 to write release notes and a newsletter for the same launch. Adopt professional LinkedIn style, enthusiastic clichés and engagement questions. Let each draft develop its own voice.

Good example

Extract one reusable profile from S1–S5, citing each trait: short concrete sentences, scoped numbers and explicit untested areas. The absence of engagement questions is an observation about these samples, not a permanent rule about the author. Apply this profile to both drafts. Launch facts: bulk export is added, only administrators can enable it, and mobile is untested. The newsletter may add context but no new facts or exaggerated effects.

Why the change matters

Platform clichés lack evidence in the author’s samples, while independently improvised drafts can diverge. A compact profile makes stylistic decisions traceable and reusable. Separating facts from style also prevents a sample’s numbers from becoming another draft’s results.

Observable expectation

An illustrative profile cites S5 for scoped numbers and S2 for explicit unknowns. Both drafts retain administrator enablement and untested mobile behavior; compare sentence rhythm with the samples.

Check every trait for evidence and reject invented launch effects. Unsupported traits are suggestions rather than author facts. Author acceptance still requires review.

Limits

Sample count and age affect representativeness; five is this fixture’s input, not a universal minimum. Use and retain private samples only within authorization. A voice profile is contextual evidence rather than a writing policy for all authors.