> ## Documentation Index
> Fetch the complete documentation index at: https://docs.puddin.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Understanding Review Signals in Puddin Admissions Review
> Learn what review signals are, how each signal type is calculated, and how to use signals to prioritise your review workload effectively.
Review signals are patterns in a submission's writing process data that may indicate it warrants closer attention during review. Puddin calculates signals automatically by analysing the document creation history captured during writing — things like how long the applicant spent typing, whether large blocks of text were pasted in, and how the document evolved across sessions. Signals are designed to help you allocate your review time more effectively, not to pass judgement on any submission.
Review signals highlight patterns for human consideration — they are not verdicts, scores, or indicators of misconduct. A signal may have a completely innocent explanation. Every submission deserves thoughtful human review, and signals should inform that review process, not replace it.
## How signals are displayed
Each submission's signals are summarised in the **Review Signals** bar at the top of the review interface. Signals are marked **High**, **Medium**, or **Low** based on how strongly the pattern appears relative to your cohort. Click **View all signals** to see the full breakdown.
You can also filter the applicant list by signal level to build a prioritised review queue — for example, showing all submissions with at least one **High** signal first.
## Signal types
**What it measures:** The proportion of the final document's content that was inserted via paste actions rather than typed character-by-character.
**How it's calculated:** Puddin totals the character count of all paste events recorded during writing and compares this to the total character count of the submitted document. A submission is flagged when pasted content makes up a significant proportion of the final text.
**What it might indicate:** The applicant copied substantial amounts of text from elsewhere — which could include drafting in another application, transferring notes, or sourcing text from another person or tool.
**What it doesn't mean:** Pasting alone is not evidence of any specific behaviour. Many applicants draft in word processors and paste their work into Puddin, which is a normal part of writing. The paste events panel lets you inspect exactly what was pasted and when, giving you the context to make an informed judgement.
**What it measures:** The total time the applicant spent actively typing in the document, relative to the length of the submitted text.
**How it's calculated:** Puddin tracks active typing periods and distinguishes them from idle time (periods where no keystrokes were recorded). Short active writing time means the document was produced with relatively little observable typing activity. This signal is calibrated against your cohort, so it reflects what's unusual for your specific applicants.
**What it might indicate:** The document may have been assembled primarily from pasted content, or written largely outside the Puddin environment before being transferred in.
**What it doesn't mean:** Fast typists may legitimately have short active writing times. This signal is most meaningful when combined with the High Paste Volume signal.
**What it measures:** Whether the applicant completed the entire personal statement in a single, uninterrupted writing session.
**How it's calculated:** Puddin flags submissions where all writing activity occurred within a single session, particularly when that session is very short relative to the document length.
**What it might indicate:** The document may have been prepared entirely before the applicant opened Puddin, then transferred in during one session rather than composed iteratively.
**What it doesn't mean:** Some applicants work intensively and produce well-crafted writing in a single sitting. Context matters — a single session of substantial duration with varied typing patterns is very different from a brief session with high paste volume.
**What it measures:** The absence of typical editing and revision behaviour across the document's creation history.
**How it's calculated:** Puddin analyses the revision timeline for patterns that characterise authentic iterative writing: rereading, editing earlier sections, adding and removing sentences, and refining phrasing over time. Submissions with very little revision activity — particularly when combined with other signals — are flagged.
**What it might indicate:** The document arrived in something close to its final form rather than being developed through the typical drafting process.
**What it doesn't mean:** Some writers produce clean first drafts and make few structural revisions. This signal is intended as one data point among many, not a standalone indicator of anything.
## Using signals to prioritise review
Puddin does not make decisions — your reviewers do. Signals are a tool to help you direct limited review time where it is most needed:
1. **Filter by signal level** — Use **Filter → Review Signals: High** to surface submissions that may need the most scrutiny.
2. **Look at the combination of signals** — A submission with multiple active signals warrants more careful review than one with a single, isolated pattern.
3. **Read the writing process evidence** — Use the paste events panel and revision history to understand the context behind any signal before drawing conclusions.
4. **Apply consistent standards** — Review all flagged submissions using the same criteria. Document your reasoning in the review notes field so decisions are traceable.
Signal thresholds are calculated relative to your cohort, so what counts as "high paste volume" in one cycle is calibrated to your own applicant pool — not a universal fixed value. This means signals remain meaningful even if your applicants as a group tend to use copy-paste more or less than average.