Turns testimonials from "hey can you write something nice" into actual sales assets with numbers and specificity. The quality scoring framework is the real value here: it won't let you ship vague praise, and it includes iteration scripts to push clients for concrete metrics when they give you generic feedback. Three modes scale from a single outreach script to a full VA-ready collection system with multi-channel campaigns. The voice matching guidance is practical, pulling from actual client emails instead of writing ad copy. If you've ever gotten back "it was great!" and didn't know how to salvage it, the iteration protocol handles that exact scenario.
npx -y skills add brianrwagner/ai-marketing-claude-code-skills --skill testimonial-collector --agent claude-codeInstalls into .claude/skills of the current project.
Detect from context or ask: "One script, full system, or full system with campaign?"
| Mode | What you get | Best for |
|---|---|---|
quick | 1 outreach script + 1 format template | Asking one client, right now |
standard | Full collection system: timing, scripts, formatting, display guidance | Building repeatable social proof |
deep | Full system + multi-channel strategy + case study pipeline | Sales enablement, proposal library |
Default: standard — use quick if they have one client in mind. Use deep if they want a system they can hand to a VA.
Before proceeding, gather:
If results are vague (e.g., "things improved"), stop and ask: "Can you name one specific number — even a rough estimate? That's what makes a testimonial credible and usable." Do not draft until you have this.
If the user wants to skip a field: note it and flag the quality impact in the output.
Before drafting anything, reason through:
Output a brief situation summary:
"You have a [length] engagement with [client] in [industry], with [strong/weak] results data. I'll draft in [format] with [authentic/templated] voice. Main gap to address: [specific gap]."
Score the raw testimonial content (or anticipated content) before drafting:
| Dimension | Score 1 | Score 3 | Score 5 |
|---|---|---|---|
| Specificity | No details | Vague references | Specific named result |
| Measurability | "It was great" | "Noticeable improvement" | "40% increase in leads" |
| Authentic Voice | Sounds like ad copy | Slightly stilted | Reads exactly how a person talks |
| Length | Too short (no context) | Decent but thin | Enough for all 3 formats |
Scoring rule:
Direct Ask:
Subject: Quick favor (30 seconds)
Hey [Name],
Loved working on [project] with you — especially seeing [specific result].
Would you be open to sharing a quick testimonial I could use on my site?
No pressure. If yes, I can either:
A) Send you 3 questions to answer
B) Write a draft for you to approve/edit
Whatever's easier.
Question Route:
3 quick questions:
1. What was the situation before we worked together?
2. What changed or improved?
3. Would you recommend this to others? Why?
Draft-on-Behalf Framework: Rules for writing in the client's voice:
Fill-in template:
"[Client situation in 1 sentence]. [What the engagement delivered — concrete].
[Specific result, ideally with a number]. [Recommendation statement in client's natural voice]."
"[One punchy outcome sentence — lead with the result]"
— [Name], [Title] at [Company]
Use for: Homepage, LinkedIn featured section, proposal proof points
"[Problem or situation]. [What changed]. [Recommendation or result]."
— [Name], [Title] at [Company]
Use for: Services page, sales decks, email sequences
Structure:
Use for: Case study pages, downloadable PDFs, high-trust sales assets
After generating all formats, evaluate:
Specificity check: Does the short version have at least one concrete outcome (not just "great results")? Voice check: Could the client have actually written this, or does it sound like a marketing headline? Placement check: Is the recommended format actually correct length for the stated use case? Ethics check: Does the draft contain any claims the client didn't make or numbers you added?
Flag any issues: "The short version lacks a specific metric — you'll need to get one number from the client before using this on a homepage."
If the received testimonial scores ≤2 on any dimension, send this gentle follow-up:
"Thanks so much — this is great. One small ask: could you add one specific
number or outcome? Even rough ('saved us about 5 hours a week') makes it
much more compelling for other clients. Totally optional, but makes a real difference."
If a second request still yields nothing specific: use Tier 3 proxy language:
"noticeable improvement in [area]" or "process now runs without manual oversight"
Always deliver a placement recommendation with the formatted testimonials:
| Format | Recommended Locations | Why |
|---|---|---|
| Short (2-liner) | Homepage, proposals, LinkedIn | Trust at first glance |
| Medium | Services page, email, sales decks | Overcome late-stage objections |
| Long | Case study page, PDF, portfolio | Deep proof for serious buyers |
Cross-reference: If this client has a strong story, suggest running case-study-builder to expand into a full case study.
## Testimonial: [Client Name] — [Date]
### Quality Assessment
- Specificity: [X/5]
- Measurability: [X/5]
- Authentic Voice: [X/5]
- Length: [X/5]
- **Total: [X/20] — [Ready to use / Needs iteration]**
### Short Format (2-liner)
"[Quote]"
— [Name], [Title], [Company]
### Medium Format
"[Quote]"
— [Name], [Title], [Company]
### Long Format
[Full narrative]
### Placement Recommendation
[Where to use each format]
### Next Step
[Iteration note OR cross-reference to case-study-builder]
Skill by Brian Wagner | AI Marketing Architect | brianrwagner.com
sickn33/antigravity-awesome-skills
moizibnyousaf/ai-agent-skills
github/awesome-copilot