Reports & Evaluation

Reports, Scoring, and Signals

How screening results are structured, how scores are calculated, what completion states mean for reporting, and how verification works.

0–100% Category Score Range

4 Levels Evaluation Guidelines

8-Char Verification Code Format

Overview

When a candidate completes a screening, the system evaluates their entire conversation against the rubric built into the Sage and generates a structured report. That report is the primary output of the screening process: a scored, explained, verifiable assessment of the candidate's performance.

Not every screen produces a full report. Only screens that reach the Finished completion state generate scored results. Screens that end early, time out, or are abandoned produce a completion record but no evaluation.

Report Structure

A completed screen with full results contains several distinct components:

Summary — A written assessment of the candidate's overall performance, generated by the evaluation engine from the full transcript.

Recommendation — A pass or do-not-pass recommendation derived from the scores and the Sage's evaluation guideline.

Category Scores — Percentage scores (0–100) for each evaluation category defined during the Sage build, each with an explanation and supporting quotes from the candidate's responses.

Overall Score — An aggregate percentage calculated as the average of all individual category scores.

Transcript — The complete conversation log between the candidate and the Sage, available for manual review.

Verification — A unique verification code and QR code for external validation.

The report is viewable on the screen's detail page and can be downloaded as a PDF. The PDF includes the summary, all category breakdowns with scores and explanations, the verification code, and a QR code in the header.

How Scoring Works

Category Scores

Categories are defined during the Sage build process, derived from your role context and cultural alignment inputs. Each category represents a distinct evaluation dimension (for example: technical competence, communication, problem-solving, or cultural fit). The specific categories vary per Sage because they are generated from your inputs, not from a fixed template.

Each category receives a percentage score from 0 to 100, accompanied by:

An explanation of why that score was assigned

Relevant quotes from the candidate's actual responses that support the assessment

Overall Score

The overall score is the arithmetic average of all category scores. It provides a single-number summary but should always be read alongside the individual breakdowns. A candidate with an 80% overall could have uniform scores across categories, or could have a 95% in one area masking a 60% in another.

The Recommendation

The system produces a pass or do-not-pass recommendation based on the evaluation guideline selected during Sage creation: Flexible, Relaxed, Balanced, or Strict. Each guideline sets a different threshold; stricter guidelines require higher performance across categories.

Guideline	Pass Threshold	Use Case
Flexible	60%	High-volume or entry-level roles
Relaxed	65%	Broad candidate pools
Balanced	75%	Standard professional roles
Strict	85%	Specialized or senior roles

The recommendation considers the overall score, individual category performance, and any configured evaluation rules. All conditions must be satisfied for a positive recommendation.

Important

The recommendation is a signal, not a verdict. Human reviewers make the final selection decision.

Completion States

A screen's completion state determines what data is available in the report. The status tells you the screen reached Completed; the completion state tells you how it got there.

State	What Happened	Results Generated
Finished	Candidate completed the full interview and submitted feedback.	Yes, full report
Ended	Candidate ended the interview early.	No
Timeout	Session exceeded the maximum duration.	No
Expired	Invitation expired during the process.	No
Abandoned	Candidate started but never returned.	No

Only Finished screens produce scored results, a summary, and a recommendation. All other states record the completion reason but generate no evaluation. The transcript is available for any screen where conversation occurred, regardless of completion state.

Integrity Signals

The system includes automated integrity checks that assess the authenticity of candidate responses. These checks produce advisory signals stored with the results and available for reviewer reference.

Signal	What It Tracks
Focus Lost	Times the candidate switched to another tab or window
Page Refreshes	Times the page was reloaded during the interview
Dev Tools	Times browser developer tools were detected open
Copy Attempts	Copy actions attempted during the interview
Paste Attempts	Paste actions blocked during the interview
AI Detection Score	0–100 score indicating likelihood that responses were human-written versus AI-generated

Note

Integrity signals do not automatically fail a candidate. Reviewers interpret them in context alongside the full transcript and make their own determination.

Verification

Every report with scored results receives a unique verification code and an associated QR code. These serve as proof of authenticity when sharing results externally.

Third parties can verify a report by scanning the QR code or entering the verification code at the verification page. Verification confirms:

The report is authentic and unmodified

The candidate's identity (name and position)

The overall score and completion date

Verification is public; the verifier does not need a SageScreen account. The verification code is generated once per screen and persists even if results are regenerated.

How to Verify a Report

Navigate to the verification page

Go to /verify or scan the QR code from the PDF report.

Enter the verification code

Enter the 8-character alphanumeric code found on the screen detail page or in the PDF report.

Enter the candidate's last name

Both the code and last name must match for verification to succeed. This two-factor approach prevents unauthorized access from someone who only has the code.

Complete the CAPTCHA and click Verify

If a CAPTCHA is enabled, complete it before submitting. The verified results will display immediately on success.

PDF Export

Results can be downloaded as a PDF from the screen detail page. The PDF includes:

Candidate name and position

Overall score and recommendation (Strong or Weak alignment)

Evaluation mode (Flexible, Relaxed, Balanced, or Strict)

Category breakdowns with scores, explanations, and candidate quotes

Executive summary

Verification code and QR code for authenticity validation

The PDF is generated on demand and suitable for sharing externally. Recipients can use the embedded verification code or QR code to confirm the report's authenticity.

Regeneration

Results can be regenerated after initial creation. This re-runs the evaluation engine against the same conversation transcript, producing fresh scores and analysis. The original results are replaced.

Regeneration is available for screens with a Finished completion state. The verification code remains the same after regeneration.

Role	Regeneration Limit
`council_member`	Once per screen
`council_power_user`	Once per screen
`council_admin`	Unlimited

Tip

If results did not generate automatically after a screen completed, use the "Generate Results Now" button on the screen detail page. If generation fails, check the error message displayed on that page for guidance.

Was this article useful?

Dislike