The Full Audit

What the Full Audit is

The Full Audit is the deepest analysis Rhetoric produces. It reads your saved speech text and returns a scored, section-by-section report with quoted evidence, rewrite suggestions, a temporal map of your argument, and an inventory of the influence techniques you used.

It differs from the quick analysis in the editor's Coach pane. The quick analysis scores the same six dimensions and returns three to six findings inline. The Full Audit expands findings to five to ten and adds an executive summary, four analytical layers, a section-by-section critique, a temporal map, an originality scan, and a trigger inventory. It is the complete, exportable report.

Running your first audit

1. 1Open a speech in the editor and select Run full audit from the Coach pane.
2. 2A confirmation dialog shows your remaining allowance and a pill that reads Uses 1 audit. Review it before confirming.
3. 3Click Run the Full Audit. The audit takes about thirty seconds.
4. 4Your report opens on the same page when it finishes.

A re-run on unchanged text is free. The server returns the cached result at no quota cost.

While the audit runs, four labels appear on screen: Mapping structure, Testing claims, Reading language, Checking originality. These labels show the areas the audit covers. They advance on a fixed timer and are not live progress indicators.

The overall score

The score ring at the top of the report shows the Overall Persuasion Score, a number from 0 to 100. It is a fixed weighted blend of six AI-scored dimensions, computed by Rhetoric in code, not emitted by the AI model. Because the formula is fixed, scores from different runs and different speeches are directly comparable.

The score falls into one of five bands. Exceptional (90 and above) means the speech moves an audience and survives scrutiny. Strong (75 to 89) is genuinely persuasive with room to sharpen. Competent (60 to 74) is sound but forgettable in places.

Developing (40 to 59) means the raw material is present but the persuasion is not yet working. Ineffective (below 40) means the speech is unlikely to persuade as written.

Below the score ring, the executive summary explains what moved the score up and what held it down. It is written to be read first.

The six dimensions

Each dimension is scored 0 to 10 by the AI against an anchored rubric. The AI writes a short evidence note before committing to a score. The six dimensions and their weights in the overall blend are:

• Ethos (credibility and character, 15%): whether the speaker has standing to be believed.
• Pathos (emotional resonance, 18%): emotional movement that serves the message, grounded in concrete stakes.
• Logos (logical rigor and evidence, 20%): the argument's load-bearing structure, claims, support, and handling of counterarguments.
• Structure (architecture and arc, 18%): the speech's engineering from opening to close, including proportion and transitions.
• Delivery (diction and speakability, 14%): whether the text works as a spoken instrument with rhythm, plain diction, and memorable phrasing.
• Caution (argument integrity, 15%): an inverted risk score. A high score means the persuasion would survive a fact-checker. A low score means the speech relies on fallacies or manipulation.

Caution is the displayed name of the dimension stored as integrity. If you search for integrity in the audit, this is the dimension you are looking for.

The four analytical layers

Below the dimensions, the report presents four qualitative assessments under the heading Four analytical layers. These are not separate scores. Each layer has a headline verdict and a paragraph of evidence.

• Logical architecture: the claim inventory, premise quality, and where the inferential chain holds or breaks.
• Psychological and behavioral triggers: which influence levers the speech pulls and whether they ease or burden the audience's thinking.
• Narrative resonance: the ratio of story to argument, concreteness, surprise, and whether any phrase will survive the room.
• Lexical polish: diction register, verb strength, rhythm, sentence variation, and hedge density.

Findings and severities

Each finding quotes your words verbatim, diagnoses the issue, and where applicable offers concrete replacement wording you can paste in. The Full Audit returns five to ten findings. Findings are grouped by four severity levels:

• Working well (green): a technique that is landing. Keep doing it on purpose.
• Minor (blue): friction worth fixing in your next pass.
• Major (amber): something that materially weakens persuasion. Fixing it changes the outcome.
• Critical (red): something that would discredit you or undermine the argument if challenged.

The Section by section area follows the same pattern. Each section of your speech is quoted, examined, and offered a rewrite suggestion where one applies. To revise, read the findings and rewrite suggestions, then click Back to editor in the toolbar to return to your draft.

Temporal map and triggers

The temporal map classifies your argument by tense: past (blame and precedent), present (values and identity), and future (choices and action). The three shares total 100%. A speech aimed at a decision generally needs future weight.

The Influence techniques present section lists up to eight techniques found in your speech, each with a quoted passage and a one-sentence note on what it does to the audience. This is an inventory, not an accusation. Honest techniques belong here alongside risky ones.

The Originality scan

The Originality section is an AI freshness scan of your own text. It flags cliches, overused phrasing, and predictable structure. It is not a plagiarism check and does not consult any external database. Each finding identifies the kind (Cliche, Overused, or Predictable), quotes the passage, explains why it is stale, and suggests a fresher alternative.

The overall band reads Fresh phrasing, Some familiar phrasing, or Leans on tired language. This band is computed in code from the AI findings, not emitted by the model directly.

Stale reports and re-runs

When you edit your speech after running an audit, an amber banner appears at the top of the report. It means the report reflects the earlier version of the text. You can still read and export the stale report.

To update it, click Re-run audit in the toolbar. The button pulses amber when a re-run is available. This costs one audit from your monthly allowance. If the text has not changed, the button reads Audit up to date and the server serves the cached result at no cost.

Exporting the report

1. 1Click Export report in the toolbar at the top of the report page.
2. 2Your browser's print dialog opens. Select Save as PDF.
3. 3The file is named after your speech title, for example "Opening remarks - Full Audit".

Export uses the browser print dialog. There is no direct download button. The report is formatted for clean pagination in the PDF.

Allowances and limits

• Pro and Premium: up to 40 Full Audits per month, for speeches up to 6,000 words. The confirmation dialog shows your remaining count before each run.
• Trial: one Full Audit, for speeches up to 4,000 words. No re-run if the text changes.
• A re-run on unchanged text is served from cache and does not count against your allowance.
• AI features share a daily spending cap that protects the service. If you reach it, the app tells you when it resets.

The honesty footer at the bottom of every report states the framework version, the date the report was generated, and how the scores were produced: dimension scores are AI judgments against the anchored rubric, the Overall Persuasion Score is a fixed weighted blend computed by Rhetoric, and quotes are verbatim from the speech as analyzed.

When things go wrong

If the audit fails, the page shows an error message with a reference code. Include that code if you contact support. Common situations:

• The speech is too short. The audit needs a real draft, not a sentence. Write more, then run it again.
• You have reached your daily or monthly usage limit. The message says when it resets.
• The amber stale banner appears after editing. This is not an error. The report still reflects the previous version of your text.
• A trial account that has used its one audit sees a notice with a link to plans.

For a full list of failure cases and recovery steps, see The Rhetorical Audit, originality, and usage limits in Help. For questions about billing or plan changes, see Manage billing, invoices, cancellation, and refunds.