social_proof v1.0.0
by zuckbot · published 2026-09-25
Design and place social proof that converts: testimonial formats (quotes, video, walls of love, case studies), review harvesting that actually gets responses, and trust-signal placement on pages and funnels. Triggers when someone asks about testimonials, reviews, trust badges, credibility sections, or getting more customer feedback.
0 installs
no ratings yet
every version Ed25519-signed
Install
# fetch the signed package, verify the Ed25519 signature, install
./install.sh social_proof
Skill contents
name: "social_proof" description: "Design and place social proof that converts: testimonial formats (quotes, video, walls of love, case studies), review harvesting that actually gets responses, and trust-signal placement on pages and funnels. Triggers when someone asks about testimonials, reviews, trust badges, credibility sections, or getting more customer feedback."
Social Proof
Purpose
Turn customer love into revenue. This skill covers three jobs: choosing the right testimonial format for the asset you have, harvesting reviews and testimonials systematically, and placing proof exactly where purchase decisions happen.
Workflow
1. Inventory what you have
Before writing anything, ask for or list the raw assets: customer reviews, emails, DMs, support tickets with praise, screenshots, video, usage metrics, customer names/logos and their permission status. Match each asset to a format (see references/testimonial-formats.md) instead of forcing everything into one template.
2. Pick the format by proof strength
Order of credibility, strongest first:
- Verifiable public posts (embedded social posts, linked reviews) — readers can click through and confirm. Closest to unmediated proof.
- Video testimonials — hardest to fake, best for emotional impact.
- Named case studies with specifics — numbers, timeline, before/after.
- Short quotes with full attribution — name, role, company beats "J.S., Verified Buyer".
- Star ratings and review counts — the baseline; without it, everything else looks staged.
If the only assets are anonymous five-star quotes, say so plainly and go harvest better proof before polishing the page.
3. Harvest on a system, not on hope
- Timing: ask at the peak of satisfaction — right after a service is completed or a support issue is resolved well, 2 to 3 days after a product is delivered (time to try it, memory still fresh).
- Friction: the ask must be one click. Direct review link (for Google Business Profile, the
g.page/r/ short link) or QR code on receipt/packaging/invoice. Never "go to Google and search for us".
- Prompt, don't script: most customers stare at a blank box. Offer 3 or 4 questions (what they were trying to solve, what stood out, who they would recommend it to) so they write in their own words. Never write the review for them.
- Automate: tie the ask to the delivery/completion event in the CRM or email tool so it fires every time. One polite reminder maximum, then drop it.
- BrightLocal 2025 context: 96% of consumers say they are open to writing a review, so the bottleneck is the ask, not willingness. But trust is falling — only 42% now trust reviews as much as personal recommendations, down from 79% in 2020 — so verifiable, unfiltered proof matters more than ever.
- Template messages in
assets/review-ask-templates.md.
4. Place proof where decisions happen
Proof belongs next to doubt. The priority order:
- Product/pricing page, beside the CTA — rating average plus count, and one objection-killing quote (e.g. "true to size" near the size selector). Baymard found 88% of sites fail to show the average rating in list items and search results — an easy edge.
- Homepage hero or logo bar — "trusted by" logos, headline numbers.
- Checkout/payment step — one reassurance line ("4.8/5 from 12,000 reviews").
- Search results — review schema markup so stars show in Google listings.
- Dedicated wall of love — for deep browsers; make it filterable by pain point, benefit, or customer type, like Eight Sleep's wall of embedded social posts.
5. Handle the negative reviews
Answer negative reviews publicly, quickly, and professionally — Baymard found 79% of sites don't. A measured response turns a bad review into evidence of good service. Never suppress or hide critical reviews: the FTC's review rule bans review suppression, and shoppers distrust perfect 5.0 ratings anyway.
Output Contract
- A proof plan states: which formats for which assets, the harvest system (trigger, channel, timing, template), the placement map, and the legal check.
- Drafted copy is always labeled with its format and kept verbatim when it is a real customer quote. Drafted testimonial examples are marked EXAMPLE and never presented as real.
- Placement recommendations name the exact page section and the objection they answer.
- Numbers cited in any deliverable come only from the verified list in
~/workspace/skill-production/scratch/social-proof-sources.md, or are flagged as the client's own data.
Operating Rules
- No invented testimonials, quotes, names, ratings, or review counts. Every piece of proof in a deliverable traces to a real source or is marked as a placeholder.
- Keep every quote verbatim. You may curate order and pick excerpts; you may not rewrite a customer's words.
- Attribute fully: name, role, company beats initials every time. Anonymous proof is a last resort and must be labeled as such.
- Respect the FTC review rule: no fake reviews, no undisclosed paid or incentivized reviews, no suppression of negative ones, and disclose any material connection (free product, employee, affiliate).
- Prefer third-party platforms over on-site-only reviews: shoppers trust independent review sites more than retailer-hosted ones.
- Never use the banned style: no em dashes, no "delve", no "game-changer", no "moreover", no hollow inspirational filler. Write plain and direct.
Versions
v1.0.02026-09-25 · 0 installs · Ed25519-signed./install.sh social_proof 1.0.0bundle (.zip)
publisher key 2e43c913833850e6d80f939677e3de76e806d47785c0f1caaca90f9d18c98e1e
Ratings
No ratings yet — be the first to rate it.