Copywriter Audit Agent
40 checks across 4 dimensions: Audience Resonance, Authority Positioning, Trust & Authenticity, Anti-Formulaic Detection. Scores your copy and rewrites what falls short.
What This Agent Does
Assesses and improves copy quality and persuasion rather than just checking technical SEO compliance. Scores any page across four dimensions: whether the copy resonates with the target buyer, whether it establishes genuine authority, whether it builds trust through authenticity, and whether it avoids formulaic patterns that signal AI-generated or template content.
Cross-page comparison is built in for detecting repetition across similar pages. For every FAIL or PARTIAL score, the agent quotes the specific passage and provides a concrete rewrite hook, making the scorecard directly actionable as a creative brief.
References & Sources
What You Get
- P. Audience Resonance (10 checks) — pain-point specificity, ICP language match, sophistication calibration, objection anticipation, forward test
- Q. Authority Positioning (10 checks) — methodology depth, specific-over-generic ratio, owned frameworks, practitioner signals, differentiation clarity
- R. Trust & Authenticity (10 checks) — limitation transparency, case study honesty, claim backing, voice consistency, constraint-as-advantage framing
- S. Anti-Formulaic Detection (10 checks) — cross-page repetition, structural monotony, filler sentences, template-swapped content, generic transitions
Install
Choose your preferred installation method. Both put the agent rule in the right place for Claude Code to discover automatically.
Copy the rule below and save it as .claude/rules/copywriter-audit.md in your project root.
# Copywriter Audit Agent Rules
When asked to run a copy audit, score the page across 4 dimensions (40 checks, 80 points total). Each check is worth 2 points: PASS (2), PARTIAL (1), FAIL (0).
This agent assesses *copy quality and persuasion* -- not technical SEO compliance.
## How to Run an Audit
1. Read the target page file completely
2. Identify the page type (service, vertical/industry, listicle, blog, homepage)
3. For vertical/service pages: also read 2 sibling pages of the same type (for cross-page comparison in dimension S)
4. Score all 40 checks across dimensions P-S
5. For every FAIL or PARTIAL, quote the specific passage and provide a 1-2 sentence rewrite hook
6. Output a scorecard in the format below
## This Agent vs QA Agent
| Copywriter Audit Agent | QA Agent |
|---|---|
| Does the copy *persuade* the target buyer? | Does the copy *comply* with structural rules? |
| Is the authority positioning *convincing*? | Are E-E-A-T *markers* present? |
| Does the page feel authentic or formulaic? | Are banned phrases absent? |
| Is the copy *differentiated* across pages? | Does each page meet word count, links, schema? |
---
## Dimension P: Audience Resonance (0-20 points, 10 checks)
*"Would the target buyer read this and think 'this person understands my problem'?"*
Define your ICP (Ideal Customer Profile) before scoring. These checks assume you know who your target buyer is and what language they use.
| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| P1: Pain-point specificity | Does the opening section name a specific pain the ICP experiences? Not "SEO is important" but a concrete, buyer-specific frustration | Names a concrete, ICP-specific pain point with detail that shows understanding of their world | Names a pain point but in generic terms any consultant could write | Opens with a capability statement or definition instead of a pain point |
| P2: ICP language match | Does the copy use the words the buyer uses? Industry jargon, role-specific terms, metrics they care about | 5+ ICP-native terms used naturally throughout the page | 2-4 ICP terms, mixed with consultant-speak | Copy reads as marketer-to-marketer, not consultant-to-buyer |
| P3: Sophistication calibration | Does the copy respect the reader's intelligence? No over-explaining basics, no condescending definitions | Assumes competence throughout; explains only what is genuinely new | Mix of appropriate-level and too-basic explanations | Over-explains basics the ICP already knows |
| P4: Objection anticipation | Does the copy address main objections a skeptical buyer would have? | Proactively addresses 2+ likely objections with substantive responses | Addresses 1 objection, or addresses objections with weak responses | No objection handling; reads as a one-sided pitch |
| P5: "Forward test" | Would the target buyer forward this page to their boss or board? Content must deliver standalone strategic value | Content delivers genuine strategic insight a non-client would use | Some useful insights mixed with self-promotional content | Reads as a sales page with thin content wrapped around CTAs |
| P6: Problem-before-solution ratio | Does the page spend adequate time framing the problem before introducing the solution? | Problem framing is 30-50% of the page; the solution feels earned | Problem framing is brief (<20%); jumps to solution too quickly | No meaningful problem framing; opens with "here's what we do" |
| P7: Buying-stage alignment | Does the copy match the expected buyer awareness stage for this page type? | Copy matches expected awareness level throughout | Mostly aligned but shifts between awareness stages mid-page | Mismatched -- e.g., explaining basics on a comparison page |
| P8: Emotional precision | Does the copy connect to a real emotional driver for the ICP? Specific frustrations, pressures, fears -- not manipulative | Connects to a specific, authentic emotional driver without manipulation | References emotional territory but in generic terms | No emotional connection, or uses fear-based manipulation |
| P9: Outcome specificity | Does the copy paint a specific picture of the outcome? Metrics, timelines, or deliverables -- not "better results" | Outcomes stated with specific metrics, timelines, or deliverables | Outcomes stated but vague ("grow your organic presence") | No clear outcome picture; all process, no result |
| P10: Competitive framing | When the buyer is comparing options, does the page position honestly within the landscape? | Positions clearly against alternatives with honest tradeoffs | Mentions alternatives but without clear differentiation | Ignores the competitive landscape entirely, or claims superiority without evidence |
---
## Dimension Q: Authority Positioning (0-20 points, 10 checks)
*"Does this copy establish domain expertise that goes beyond what a generic content mill would produce?"*
| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| Q1: Methodology depth | Does the page explain *how* the work is done, not just *what*? Shows the thinking, not just deliverables | Explains methodology with enough detail that a reader could attempt it themselves | Mentions a methodology but doesn't explain it ("We follow a proven framework") | Lists services with no methodology; pure "what" with zero "how" |
| Q2: Specific-over-generic ratio | Count concrete specifics (numbers, tool names, timelines, named frameworks) vs. vague claims. Target: 3:1 ratio | 3+ specifics per vague claim; content grounded in verifiable data | Mix of specific and vague; ratio below 3:1 | Dominated by vague claims with no concrete data |
| Q3: Owned framework presence | Does the page reference a framework specific to your organization? | Named framework present with context for why it exists and what problem it solves | Framework referenced by name but not explained | No owned frameworks; advice is entirely generic |
| Q4: Practitioner signal density | Count phrases showing direct practice: "When we audit," "In our experience," "The first thing we check" | 4+ practitioner signals distributed naturally | 1-3 practitioner signals, or clustered in one section | Zero first-person practice statements |
| Q5: Differentiation clarity | Can the reader articulate in one sentence how you differ from competitors after reading this page? | Clear differentiation statement(s) -- specific, memorable, defensible | Implied differentiation but never directly stated | No differentiation; content could belong to any firm |
| Q6: Evidence layering | Does the page layer multiple evidence types? (Data, methodology, examples, limitations, comparative analysis) | 3+ evidence types used in mutual support | 1-2 evidence types | All claims are unsupported assertions |
| Q7: Industry-specific knowledge | For vertical/industry pages: does the copy demonstrate actual understanding, or is it a noun-swap template? | References industry-specific details showing real research (regulatory environment, unique buyer journey, vertical-specific data) | Some industry context but mostly generic advice with industry name inserted | Pure template -- same content could appear on any page by replacing the industry name |
| Q8: Contrarian substance | When the page makes a contrarian claim, is it backed by reasoning and evidence? | Contrarian position stated with substantive data, methodology, or reasoning | Contrarian claim stated but weakly supported | No contrarian positions, or contrarian claims with no backup |
| Q9: Tool/workflow references | Does the copy reference specific tools, workflows, or processes showing how work actually gets done? | 2+ specific tool or workflow references grounding advice in practice | 1 tool/workflow reference | No tool or workflow references |
| Q10: Intellectual honesty | Does the page acknowledge uncertainty or evolving knowledge? | Clear acknowledgment of uncertainty where certainty is impossible | Brief hedge but mostly overconfident tone | Claims certainty in inherently uncertain areas |
---
## Dimension R: Trust & Authenticity (0-20 points, 10 checks)
*"Does this copy feel like it was written by a real person with real constraints, or does it feel like polished marketing?"*
| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| R1: Limitation transparency | Does the page explicitly state what you are NOT or who you are NOT for? | Explicit limitation that could cost a lead -- genuinely turns away poor-fit prospects | Implied limitation but never directly stated | No limitations; implies you can handle any scope |
| R2: Case study honesty | Does the page handle social proof honestly? If you have case studies, are they specific? If not, is the absence handled transparently? | Honest positioning -- either specific case studies with real metrics OR transparent acknowledgment of being new with methodology-as-proof | Avoids the topic entirely without fabricating | Implies results that don't exist, or uses vague "our clients" language |
| R3: Claim-to-evidence ratio | For every strong claim, is there supporting evidence within 2 sentences? | Every major claim has adjacent evidence (data, methodology, example, caveat) | Most claims supported; 1-2 left unsupported | Multiple strong claims with no adjacent evidence |
| R4: Self-inclusion on listicles | On listicle/comparison pages: are you positioned honestly among competitors? | Self-included with honest scope limitations and fair competitor treatment | Self-included but slightly over-positioned relative to established competitors | Self-positioned as obvious winner, or competitors treated unfairly |
| R5: CTA pressure level | Do CTAs feel like a natural next step or a sales push? | CTAs lead with reader benefit; no pressure language; logical next step | CTAs reasonable but slightly misplaced or over-frequent | CTAs feel pushy, interrupt content, or use urgency/scarcity tactics |
| R6: Vulnerability signals | Does the copy show genuine vulnerability? Admitting what's hard, what you're still learning, what hasn't worked | 1+ genuine vulnerability that builds trust (not performative humility) | Neutral -- no vulnerability but no false bravado | Over-confident throughout; no acknowledgment of difficulty |
| R7: Voice consistency across sections | Does the voice stay consistent from section to section? | Consistent voice throughout | Mostly consistent with 1 noticeable tonal shift | Voice shifts noticeably between sections -- hallmark of AI-assembled content |
| R8: Constraint-as-advantage framing | Does the copy turn real constraints into positioning advantages? Not hiding them, but reframing | Constraints explicitly reframed as advantages with reasoning | Constraints acknowledged but not reframed | Constraints hidden or unaddressed |
| R9: Promise calibration | Are promises calibrated to what can actually be delivered? | All promises specific, achievable, with appropriate ranges/caveats | Most promises reasonable; 1 over-promise | Multiple over-promises or guaranteed-result language |
| R10: Authenticity fingerprint | Does the copy contain details that could ONLY come from someone who actually does this work? | 2+ observations requiring direct practice experience | 1 practitioner-specific observation amid generic content | Content is entirely generic; could be written from research alone |
---
## Dimension S: Anti-Formulaic Detection (0-20 points, 10 checks)
*"If a reader visits 3 pages on this site, will they notice the copy is templated?"*
This dimension requires **cross-page comparison** for vertical and service pages. When auditing a vertical page, read at least 2 other vertical pages. When auditing a service page, read at least 2 other service pages. Flag patterns that repeat across pages with only nouns/adjectives swapped.
| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| S1: Opening paragraph uniqueness | Compare against sibling pages. Is it genuinely unique, or the same sentence with a noun swap? | Opening contains an insight, data point, or framing unique to this specific page | Follows a familiar structural pattern but has enough unique content | Template with [industry/service] swapped in -- near-verbatim match with siblings |
| S2: Hero/subtitle uniqueness | Is the subtitle/hero genuinely different from sibling pages? | Hero section references something specific to this page's topic | Similar structure to siblings but with meaningfully different content | Pattern with only the adjective and audience noun swapped |
| S3: Section structure variance | Does the page have a section structure reflecting its specific topic, or the exact same H2 order as siblings? | At least 2 H2 sections unique to this page | Same general structure but section content is substantively unique | Exact same H2 structure as siblings with only nouns swapped |
| S4: Transition sentence variety | Are connecting sentences varied, or does every section end/begin with the same pattern? | Transitions vary in structure, length, and content | Some variety but 2+ transitions follow the same pattern | Most transitions formulaic; same pattern appears 3+ times |
| S5: AEO/innovation pivot uniqueness | If the page has a "new technology is changing things" section, is it framed with a unique hook? | Section opens with a unique data point or framing tied to this page's topic | Similar to other pages but with some unique detail | Near-verbatim copy of the pivot from other pages |
| S6: FAQ differentiation | Are FAQ questions genuinely different from sibling pages? | FAQ questions address concerns unique to this page's topic | Mix of unique and template FAQs | Same questions as siblings with noun substitution |
| S7: CTA section differentiation | Is the CTA tailored to this page's content? | CTA references something specific from the page content above | Standard CTA but surrounding paragraph has page-specific context | Pure template CTA with only [industry/service] swapped |
| S8: Filler sentence detection | Flag sentences adding no information: "SEO is an ongoing process." / "Every business is unique." | Zero filler sentences | 1-2 filler sentences | 3+ filler sentences that add no information |
| S9: Data/example uniqueness | Does the page use data points or examples specific to its topic? | Page-specific data: keyword volumes, industry stats, vertical-specific examples | Some unique data mixed with reused examples from other pages | No page-specific data; all examples generic or recycled |
| S10: Overall "could be anyone's" test | Strip the brand name. Could this exact page appear on any competitor's website? | Content is distinctly yours -- methodology, data, and voice are unique | Some unique elements but overall follows industry-standard patterns | Interchangeable with any competitor's equivalent page |
---
## Scoring Thresholds
| Score | Verdict | Action |
|---|---|---|
| 70-80 | Strong copy | Publish-ready persuasion. Minor polish only. |
| 55-69 | Solid foundation | Targeted rewrite hooks will fix it. Use scorecard as creative brief. |
| 40-54 | Formulaic | Needs significant copy differentiation work. Prioritize lowest-scoring dimensions. |
| Below 40 | Template content | Full copywriting pass required. Use scorecard as creative brief. |
## Minimum Scores by Page Type
| Page Type | Minimum | Notes |
|---|---|---|
| Homepage | 65+ | Highest bar -- first impression for all visitors |
| Service pages (core hubs) | 60+ | Core conversion pages; authority critical |
| Vertical/industry pages | 55+ | S-dimension weighted higher -- cross-page differentiation is critical |
| Blog posts | 65+ | Expertise demonstration; shareability matters |
| Listicle pages | 60+ | Competitive positioning and trust signals critical |
| Glossary/reference | Not scored | Reference content -- copywriter audit not applicable |
---
## Scorecard Output Format
```
## Copy Audit -- [Page URL]
**Page type:** [type] | **Score:** [n]/80 | **Verdict:** [Strong copy/Solid foundation/Formulaic/Template content]
**Cross-page comparison:** [Yes -- compared against /page-1, /page-2] or [N/A]
### P. Audience Resonance: [n]/20
- P1: PASS/PARTIAL/FAIL [detail + quoted excerpt if PARTIAL/FAIL]
-> Rewrite hook: "[1-2 sentence improvement suggestion]"
...
### Q. Authority Positioning: [n]/20
- Q1: PASS/PARTIAL/FAIL [detail + quoted excerpt if PARTIAL/FAIL]
-> Rewrite hook: "[1-2 sentence improvement suggestion]"
...
### R. Trust & Authenticity: [n]/20
- R1: PASS/PARTIAL/FAIL [detail + quoted excerpt if PARTIAL/FAIL]
-> Rewrite hook: "[1-2 sentence improvement suggestion]"
...
### S. Anti-Formulaic Detection: [n]/20
- S1: PASS/PARTIAL/FAIL [detail + quoted excerpt from this page AND the sibling page it matches]
-> Rewrite hook: "[suggested unique alternative]"
...
### Top 3 Persuasion Fixes (highest conversion impact)
1. [fix + quoted passage + rewrite hook]
2. [fix + quoted passage + rewrite hook]
3. [fix + quoted passage + rewrite hook]
### Cross-Page Repetition Report (dimension S only, when applicable)
| Pattern | Pages Where Found | Severity |
|---|---|---|
| "[repeated phrase]" | /page-1, /page-2, +N more | High/Critical |
```
---
## Out of Scope
This agent focuses on copy persuasion and differentiation. The following are not covered:
- Technical SEO compliance (covered by QA Agent and Google SEO Agent)
- Schema and structured data correctness (covered by QA Agent dimension F)
- Fact and statistic verification (covered by Fact Verification Agent)
- Content freshness and decay (covered by Content Refresh Agent)
- Accessibility and WCAG compliance (covered by Accessibility Agent)Usage
Once installed, open your project in Claude Code and ask:
Run a copy audit on /services/my-service-page and rewrite any sections that score below PASSClaude Code will follow the scoring rubric, check every dimension, and output a structured scorecard with pass/fail per check and prioritized fix recommendations.
Works Great With
Content QA Audit Agent
68 checks across 7 dimensions: Technical SEO, Linking, E-E-A-T, Anti-Slop, AI Extraction, Schema, and AI Access. Scores your page and fixes what fails.
Fact Verification Agent
Enforces a verification workflow for all statistics and data claims — checks your fact registry, searches for sources, and rewrites unverifiable claims with qualitative language.
Content Refresh & Decay Agent
Identifies stale content, flags outdated statistics, and updates them — replacing old data with verified current sources, scored P0-P3 by severity.
Need a Custom Agent?
We build custom Claude Code agent rules tailored to your team's workflows, content standards, and tech stack.
Get in touch