Writprint
Specimen
Not for filing
Sample ReportFile DEMO/B-002

An anonymous complaint to HR

A mid-sized company received an anonymous email to the HR alias alleging a hostile pattern of behavior from a senior manager. The message could only have come from one of four people on the manager's team based on the specific incidents referenced.

Inputs4 candidates
01Reporter A

"hey just wanted to flag — the standup ran long again today and we didn't get to the prod incident discussion. can we look at trimming the round-robin? feels like we're spending the first 20 mins on stuff that could be a slack thread. happy to draft a proposal if useful."

02Reporter B

"Following up on our 1:1 — I drafted the Q2 plan with a few revisions per your feedback. Three things I'd flag: (1) the staffing assumption depends on the contractor renewal, which is still TBD; (2) we should align with Marketing on the launch window before locking dates; (3) I'd like 30 minutes this week to discuss the OKR weighting. Available Thursday after 2."

03Reporter C

"I'm sorry but this is genuinely not okay. we cannot keep moving deadlines on a team this small without acknowledging what it does to people. I've watched two of my reports start looking, and frankly I don't blame them. when can we actually talk about this."

04Reporter D

"Per the discussion in the leadership sync, the roadmap pivot is operationally feasible but creates downstream coordination cost with the Platform team. I would recommend we formalize the dependency with a written agreement before public commitment. Happy to author the first draft if delegated."

Target document (anonymous)

"I want to be clear that I'm not writing this lightly. Over the last six months I've watched [redacted] do things that I genuinely do not think a person in his role should be doing. The pattern is not subtle. It is also not, in any one instance, the kind of thing that rises to a clear policy violation, which is exactly why it has gone on. I am tired of pretending it hasn't. I have asked twice in 1:1s whether anyone else has noticed. The answer is always some version of, well, that's just how he is. I do not accept that."

Determination
Reporter C
Confidence
medium
Method
Closed-set stylometric attribution
with LLM-assisted feature extraction
Executive summary

The target is most consistent with Reporter C across all measured stylistic dimensions, with strong positive matches on first-person framing, intensifier usage, register, and rhetorical structure. The signal is meaningful but not conclusive given the short sample length and the tendency of high-emotion writing to compress individual style. Recommend treating this as a strong directional indicator pending corroboration through other investigative means.

Ranked Attribution4 candidates
01

Reporter C

Primary attribution
71%

The target shares Reporter C's most distinctive features: emotional directness with first-person framing, sentence-final intensifiers ("I've watched," "I am tired"), and a willingness to use unmoderated declarative judgments ("I do not accept that"). Both samples open with a self-justifying statement of stakes ("I'm not writing this lightly" / "I'm sorry but this is genuinely not okay") before moving to a specific complaint.

Stylistic Evidence
  • First-person singular pronouns per 100 words: target ≈ 9.0, Reporter C ≈ 8.2, others ≤ 1.5
  • Sentence-initial "I" frequency: target 5×, Reporter C 4× in 50 words, others 0–1×
  • Use of intensifiers "genuinely," "actually," "frankly" — present in both target and Reporter C, absent in others
  • Both samples deploy a rhetorical concession before escalation ("I am tired of pretending" / "I don't blame them")
  • Both avoid corporate registers ("per the discussion," "flag," "OKR") that dominate Reporters A, B, and D
02

Reporter A

12%

Reporter A's sample has direct language and willingness to flag friction, but the register is consistently lowercase-informal ("hey just wanted to flag," "can we look at") with frequent use of softening hedges ("feels like," "happy to draft") that do not appear in the target. The target's escalation tone is not characteristic.

Stylistic Evidence
  • Reporter A uses lowercase sentence-initial casing throughout — absent in target
  • Hedges ("feels like," "happy to," "could be") frequent in sample, absent in target
  • Reporter A uses no emotional intensifiers; target uses several
03

Reporter D

9%

Reporter D's writing is procedural and management-coded ("per the discussion," "operationally feasible," "formalize the dependency"). The target document contains none of these markers and operates in a fundamentally different register.

Stylistic Evidence
  • Corporate-register vocabulary in sample, absent in target
  • Sample uses passive voice and nominalization heavily; target is direct active voice
  • No first-person emotional framing in sample; pervasive in target
04

Reporter B

8%

Reporter B's sample is structured and enumerated ((1), (2), (3)) with explicit calendar coordination. The target is unstructured prose with no enumeration, no logistics, and no concrete asks. Stylistic profiles diverge across nearly every measured dimension.

Stylistic Evidence
  • Enumerated lists in sample, absent in target
  • Concrete logistics ("Thursday after 2") in sample, absent in target
  • Sentence-final periods only; target uses sentence-final periods plus declarative judgments
Limitations and caveats
  • Samples are short (~50 words each). Above 300 words per sample, signal strength roughly doubles.
  • Anonymous complaints often involve deliberate stylistic masking. Compared sample writing was likely written without that intent.
  • Emotional content compresses individual style — when people are upset, they often write more like one another than like themselves at baseline.
  • Do not confront the suspected author on the basis of this report alone. Use as one input in a broader investigation.
This report constitutes investigative analysis only and is not admissible as standalone evidence in any judicial or quasi-judicial proceeding. Stylometric attribution carries irreducible uncertainty even at high confidence. The reader is advised to corroborate findings through other investigative means before acting on them.
Bring us a real case

Five design partners. Free pilot.

Try it free →