trust-credibility-and-social-proof

name	trust-credibility-and-social-proof
description	Use when selecting, verifying, and placing testimonials, reviews, logos, ratings, usage numbers, certifications, guarantees, security cues, awards, team credentials, or case-study proof. Use landing-page-and-conversion-design to compose the whole page and design-ethics-and-anti-dark-patterns for persuasion review.
metadata	{"portable":true,"category":"14-conversion-and-web-page-patterns","compatible_with":["claude-code","codex"]}

Trust, Credibility & Social Proof

Acknowledgement: Shared by Peter Bamuhigire, techguypeter.com.

Proof only works when it is believed. Fabricated or anonymous proof, once suspected, drives trust below having no proof at all. This skill designs the trust/proof section as an authored, honest instrument: real people named and attributed, security and privacy cues that are true, a credibility hierarchy that leads with the strongest real signal you have, and not a single manufactured one. Per doctrine/design-doctrine.md §0, the moat is looking authored and trustworthy by skilled human hands — a faked trust badge reads as exactly the opposite the moment it is noticed, and it is noticed.

Use When

• You are designing the trust / social-proof / credibility section in isolation — the testimonials block, the customer logo strip, the ratings/usage row, the security-and-privacy cues, the certifications/press/awards strip, or a case-study proof panel.
• A page converts poorly because the visitor doesn't believe the claim — the value prop is clear but there is a credibility gap, and you need to choose and arrange the right proof to close it.
• You have real proof to deploy (named testimonials, true metrics, genuine logos, real certifications) and need the credibility hierarchy: what to lead with, what to subordinate, where each signal earns placement.
• Proof is genuinely thin (a new product) and you need the honest substitutes — founder credibility, transparent methodology, a real guarantee — instead of fabricating proof.
• Someone is reaching for a fake review count, a borrowed logo, an anonymous "★★★★★", a "trusted by millions" with no source, or "10,000+ users" you can't back — STOP and use the authenticity rules here.

Do Not Use When

• You are building the whole landing/marketing page (hero, value prop, mechanism, objections, CTA hierarchy) — use 14-conversion-and-web-page-patterns/landing-page-and-conversion-design. That skill composes this proof section into the page spine; this skill details the proof section itself.
• The task is the ethics gate / dark-pattern audit across the whole experience (fake scarcity, confirmshaming, forced continuity, roach-motel) — use 00-cross-cutting-ops-qa-a11y/design-ethics-and-anti-dark-patterns. This skill stays within the proof section and defers the cross-cutting ethics catalog there; it only forbids the proof-specific dark patterns (fabricated/anonymous proof, fake counts, fake scarcity).
• You only need the wording of a testimonial pull-quote or a badge label — use 10-content-design-and-ux-writing/ux-writing-and-microcopy, then return here for selection and placement.
• The proof is a data exhibit / KPI dashboard (charts of real metrics) — use 12-data-viz-and-dashboards. This skill handles proof as persuasion in a marketing context, not analytical dashboards.

Required Inputs

Input	Source	Required?	Evidence
Audience anxieties, claims, objections, and decision stage	Research, sales, and support	yes	Prioritised trust questions
Candidate proof and attribution rights	Customers, legal, security, and brand owners	yes	Source, date, scope, and permission
Page context and conversion goal	landing-page-and-conversion-design or product owner	yes	Placement brief

• The credibility gap. The specific doubt this section must answer: "is it real?", "is it for someone like me?", "is it safe with my data?", "will it actually deliver?" Different doubts need different signals — name the doubt before choosing the proof.
• The real proof inventory you are allowed to use. A list of what genuinely exists and is cleared: which customers will be named, which quotes are real and attributed, which metrics are defensible and internally sourced, which certifications are actually held, which logos belong to paying customers. If a logo isn't a real customer or a metric can't be backed, it is out.
• Who the visitor is. Proof from someone like the visitor (same role, sector, size) out-converts a famous-but-irrelevant name. You cannot match the proof without knowing the audience.
• The privacy/security posture when relevant — what certifications, encryption, data-residency, and compliance facts are true and current (not aspirational), so security cues are honest.

0. Name the doubt, then pick the proof type that answers it

Map the credibility gap (above) to the signal class in references/trust-signal-catalog.md. "Is it real / do others use it?" → social proof (testimonials, logos, usage numbers). "Is it for someone like me?" → matched testimonials + named case studies. "Is my data safe?" → security/privacy cues. "Will it deliver?" → guarantees, results metrics, certifications. Don't spray every badge — deploy the signal that answers this visitor's actual doubt.

1. Build the credibility hierarchy — lead with your strongest REAL signal

Order proof by persuasive weight for this audience, strongest first (full ranking in references/trust-signal-catalog.md §Credibility hierarchy):

1. A named, attributed testimonial from someone like the visitor, with a result — the single strongest signal. Name + role + company + (face when permitted) + a concrete outcome.
2. A real, defensible aggregate number — "2,400 finance teams", "average close time fell 9.1 → 2.4 days" — stated precisely and internally sourced.
3. Genuine customer logos (only paying customers) and third-party ratings (real G2/Capterra score with the real review count linked).
4. Security/privacy cues when data-safety is the doubt — true certifications (SOC 2 Type II, ISO 27001), encryption-in-transit/at-rest facts, GDPR posture, each linking to evidence.
5. Founder/team credibility, press, awards, certifications — supporting, not leading, unless the audience buys on authority (regulated sectors, high-consideration B2B).

Lead with #1 if you have it; fall back down the list to the strongest signal you can honestly field. Never promote a fabricated signal up the hierarchy to fill a gap.

2. Make every testimonial authentic and attributed (the non-negotiable)

Per references/trust-signal-catalog.md §Authenticity and the engine ethics posture:

• Attribution is mandatory — name + role + company, a face where the person consented. An anonymous "★★★★★ — A happy customer" reads as invented and lowers trust below no testimonial.
• Never fabricate a testimonial, a rating, a logo, a review count, or a metric. Using the logo of a company that is not a customer is dishonest and legally exposed (passing-off / false endorsement).
• Real numbers, stated precisely, internally sourced and defensible. "Reduced support tickets 34%" beats "dramatically reduced tickets" — and you must be able to back it.
• Disclose paid/affiliate relationships (FTC and equivalents). Plain disclosure increases credibility; hidden sponsorship destroys it.
• Show, don't just claim where possible — a real screenshot, a named case study with numbers, a verifiable rating link beats an adjective.

3. Make security & privacy cues honest and verifiable

A trust badge that doesn't link to evidence is decoration, and a badge for a certification you don't hold is fraud. Per references/trust-signal-catalog.md §Security & privacy:

• Show only true, current certifications/compliance (SOC 2, ISO 27001, HIPAA, PCI-DSS, GDPR) — each linking to evidence (the report request, the certificate, the policy), not a static image.
• State concrete data facts where they reassure: encryption in transit and at rest, data residency, the cancellation/data-deletion path. Vague "bank-level security" with nothing behind it is the security-cue equivalent of an anonymous testimonial.
• Place security cues where the doubt arises — next to the form, the payment step, the data-import CTA — not buried in a footer if the data fear is the conversion blocker.

4. When proof is genuinely thin — honest substitutes, never fabrication

A new product has no testimonials yet. The honest moves (full list in references/trust-signal-catalog.md §Thin-proof):

• Lead with founder/team credibility (real bio, real track record) and transparent methodology ("here is exactly how it works").
• Offer a real, strong guarantee or a genuine free tier — honest risk reversal substitutes for proof you don't have.
• Use real early-stage framing if true — "join the first 100", "early access" — which is honest and can itself be attractive.
• Never invent a testimonial, borrow a logo, or post a fake review count to fill the gap. The honest thin-proof section out-converts the padded one that gets caught.

5. Hold accessibility and performance — proof is content, not decoration

• Accessibility (doctrine/references/wcag-2.2-criteria.md): testimonial faces need real alt text (or alt="" if purely decorative and the name is in text); a star rating must not be colour-only — pair it with the numeric value and a text label (SC 1.4.1); badge contrast meets the UI-component floor (3:1); any "verified review" link is a real focusable control with a visible focus ring. Don't lock proof behind hover (interaction-anti-patterns E3).
• Performance (doctrine/references/web-performance-budgets-2026.md): logo strips and headshots are image weight — serve sized, compressed AVIF/WebP with width/height set to reserve space (protect CLS); lazy-load below-the-fold proof; don't pull a heavy third-party review widget that tanks INP.

6. Distinctiveness — the proof section is a slop magnet too

Run 04-web-and-ui-design/distinctive-by-design thinking here as well: the convergent default is a centred "Loved by teams everywhere" heading over a grey logo strip and three identical quote cards with round avatars. Authored proof looks specific — real faces, a quoted result pulled large, an asymmetric layout, the brand's own type carrying the pull-quote. Avoid the templated three-equal-card testimonial row as a reflex (per doctrine/references/interaction-anti-patterns.md D1 flat hierarchy: the result in the quote should be the largest thing, not the avatar).

7. Specify, then build

Produce a proof-section spec (see examples/credibility-section-spec.md): the named doubt, the chosen signals in hierarchy order, the real attributed content for each, the placement, the accessibility/perf treatment, and the authenticity checklist completed.

Workflow

1. Rank the audience's trust questions and objections by decision risk.
2. Inventory candidate proof with source, scope, date, permission, and expiry.
3. Match the strongest relevant proof to the claim and point of hesitation.
4. Remove unverifiable, misleading, outdated, or over-broad signals.
5. Review placement, accessibility, legal scope, and expiry ownership before publication.

Authenticity & Trust Checklist (run before building)

1. The doubt is named and each signal chosen answers it (no badge-spray).
2. Credibility hierarchy leads with the strongest real signal; nothing fabricated is promoted up.
3. Every testimonial is attributed — name + role + company (+ face with consent); zero anonymous "★★★★★".
4. Zero fabrication — no invented quotes, no borrowed/non-customer logos, no fake review counts, no unbackable metrics, no "trusted by millions" without a source.
5. All numbers are precise and internally defensible — and you can produce the source.
6. Security/privacy cues are true, current, and link to evidence; placed where the doubt arises.
7. Paid/affiliate relationships disclosed (FTC and equivalents).
8. No fake scarcity/urgency anywhere in the proof section ("only 2 left", "3 viewing now", resetting countdowns) — defers to design-ethics-and-anti-dark-patterns; never fabricated here.
9. Accessibility — ratings not colour-only, faces have correct alt, badges meet 3:1, links focusable, proof not hover-locked.
10. Performance — proof images sized/compressed/CLS-reserved; no INP-tanking review widget.
11. No convergent default — not a centred grey logo strip + three identical avatar cards by reflex.

If any item can't be satisfied (proof is thin, a stakeholder wants a borrowed logo or a fake count), say so and ask — never quietly ship fabricated or anonymous proof.

Decision Rules

Condition	Proof choice	Wrong-choice failure
Verifiable outcome proof exists	Lead with specific result, context, and attribution	Generic praise fails to reduce risk
Certification scope is current and relevant	State exact scope and verification path	Badge implies broader compliance than earned
Proof is thin	Use transparent process, guarantees, or founder credibility	Fabrication creates ethical and legal exposure

Capability Contract

Read and search are required for proof sources, dates, permissions, and page context. Editing is allowed only for authorised composition. Contacting customers, using logos, publishing claims, or asserting certification requires separate authority.

Degraded Mode

Without attributable proof, produce a proof-gap register and safe placement plan. Stop publication of testimonials, logos, ratings, counts, awards, or compliance claims until permission, scope, date, and source are verified.

Anti-Patterns

• Anonymous testimonials — "★★★★★ — A happy customer", a quote with no name/role/company. Reads as invented; lowers trust below no testimonial.
• Borrowed / non-customer logos — a logo strip of companies that don't actually use the product. Dishonest and legally exposed.
• Fabricated or unbackable numbers — "trusted by millions", "10,000+ users", "★★★★★ (4,212 reviews)" you can't source. One challenged number poisons every other claim on the page.
• Decorative trust badges — a SOC 2 / "secure" image that links nowhere, or a badge for a certification not actually held. Decoration at best, fraud at worst.
• Fake scarcity/urgency masquerading as social proof — "3 people are viewing this", "only 2 left" that resets on reload, a countdown that restarts each visit. Banned; route honest equivalents via design-ethics-and-anti-dark-patterns.
• Avatar bigger than the result — the round headshot is the largest element and the actual outcome is tiny caption text (flat-hierarchy failure, interaction-anti-patterns D1).
• Badge-spray — every certification and award dumped indiscriminately, answering no specific doubt and reading as insecure over-justification.
• Security cues buried in the footer when data-safety is the actual conversion blocker — place the cue where the fear arises (the form, the import step).
• The convergent template — centred "Loved by teams everywhere" + grey logo strip + three equal avatar cards. The reflexive proof section; the Mission targets it.

Outputs

Artefact	Consumer	Evidence and acceptance condition
Trust-question and proof hierarchy	Marketing and product teams	Strongest evidence answers the highest-risk objection
Attributed proof inventory and placement map	Designers and writers	Each signal has source, scope, date, permission, and destination
Verification and expiry record	Legal, security, and release owner	Claims remain current, accurately scoped, and removable on expiry

• A named credibility gap and the proof signals chosen to close it, in hierarchy order.
• A proof-section spec (per signal: the real attributed content, placement, type/space treatment, the accessibility and performance treatment) in the shape of examples/credibility-section-spec.md.
• A completed authenticity & trust checklist confirming every signal is real and attributed, with zero fabricated, anonymous, or fake-scarcity proof.

Examples

• examples/credibility-section-spec.md — a full, real proof-section spec for a sample product (a small-clinic patient-records SaaS), showing the named doubt, the credibility hierarchy applied, real attributed testimonials, honest security/privacy cues placed at the point of fear, the thin-proof fallback discussion, and the WCAG/perf constraints — with the authenticity checklist completed.

References

• references/trust-signal-catalog.md — the full catalog of trust signals (social proof, security & privacy cues, authority/credentials, guarantees, transparency), the credibility hierarchy (what to lead with), the authenticity rules, the thin-proof substitutes, and the placement guidance.
• doctrine/design-doctrine.md — the Mission (§0, "the moat is looking human-made and trustworthy"): fabricated proof reads as the opposite of authored the moment it is caught.
• doctrine/references/interaction-anti-patterns.md — D1 flat hierarchy (the result, not the avatar, is the focal element), D3 colour-only signalling (ratings), E3 hover-dependency (don't hover-lock proof).
• doctrine/references/wcag-2.2-criteria.md — non-colour rating state, image alt, focusable verified-review links, badge contrast.
• doctrine/references/web-performance-budgets-2026.md — proof images sized/compressed/CLS-reserved; avoid INP-tanking third-party review widgets.
• Sibling skills: 14-conversion-and-web-page-patterns/landing-page-and-conversion-design (composes this proof section into the whole page; defers the testimonial/security-cue detail here), 00-cross-cutting-ops-qa-a11y/design-ethics-and-anti-dark-patterns (the cross-cutting dark-pattern catalog — no fake scarcity/urgency; this skill defers there and forbids only proof-specific fakery), 04-web-and-ui-design/distinctive-by-design (the proof section is a slop magnet too), 10-content-design-and-ux-writing/ux-writing-and-microcopy (testimonial pull-quote and badge wording).