Referral Engine

Design a customer referral program with incentive structures, sharing mechanics, fraud prevention rules, and tracking setup that turns existing buyers into a...

Security Scan

The skill is an instruction-only design assistant for referral programs and its inputs and behavior align with that purpose; it requests no credentials, performs no installs, and contains no unexpected instructions.

Purpose & Capability

The name and description match the SKILL.md content: it asks for business metrics (AOV, LTV, sales channels, tech stack) and promises a blueprint for referral programs. There are no unrelated requirements (no cloud keys, no platform-specific credentials).

Instruction Scope

The SKILL.md limits runtime behavior to analyzing user-provided inputs and producing a referral program blueprint. It does not instruct reading system files, accessing environment variables, calling external endpoints, or exfiltrating data.

Install Mechanism

There is no install spec and no code files — this is instruction-only, so nothing will be written to disk or downloaded during install.

Credentials

The skill requires no environment variables, credentials, or config paths. All requested inputs are user-provided business metrics and optional stack details, which are proportionate to the stated purpose.

Persistence & Privilege

The skill does not request always-on presence and uses normal model invocation settings. There is no attempt to modify other skills or system-wide settings.

Assessment

This skill is low-risk: it only produces a written referral program blueprint and does not install code or ask for credentials. Before using, avoid pasting real customer PII or secrets into prompts; validate any fraud-prevention or tracking recommendations with your legal/privacy teams (cookie consent, data-retention rules), test proposed rules in a limited rollout to avoid customer disruption, and ensure any implementation steps are applied only in trusted referral/tracking tools (ReferralCandy, Shopify apps, analytics) rather than copying unverified code directly into production.

Design a complete customer referral program that transforms your existing buyer base into a reliable, low-cost acquisition channel. This skill walks you through building incentive structures, defining sharing mechanics across platforms, setting up fraud prevention guardrails, and establishing tracking infrastructure so every referral dollar is accountable and every new customer is attributed correctly.

Use when

• You want to launch a refer-a-friend program for your Shopify, TikTok Shop, or Amazon storefront and need a structured plan covering incentives, rules, and tracking
• A founder or growth manager says "we need our customers to bring us more customers" and you need to design the full referral loop from scratch
• You are evaluating whether to offer cash-back, store credit, percentage discounts, or free products as referral rewards and need a framework to decide
• Your existing referral program has low participation or high fraud rates and you need to redesign the incentive structure and add abuse prevention rules

What this skill does

This skill analyzes your product type, average order value, customer lifetime value, and existing marketing channels to design a referral program tailored to your business. It determines the optimal reward type and amount for both the referrer and the referred friend, maps out the sharing flow across email, SMS, social media, and unique referral links, defines fraud prevention rules such as self-referral blocks, IP deduplication, minimum purchase requirements, and velocity limits, and produces a tracking plan covering attribution windows, conversion events, and reporting dashboards. The output is a ready-to-implement blueprint that balances generosity with profitability.

Inputs required

• Product category and AOV (required): What you sell and the average order value, so reward sizing is proportional to margin — e.g., "skincare, AOV $45"
• Estimated customer LTV (required): Rough lifetime value per customer so the skill can set a reward ceiling that keeps CAC below LTV — e.g., "$120 over 12 months"
• Sales channels (required): Where you sell (Shopify storefront, TikTok Shop, Amazon, retail) so sharing mechanics and tracking are channel-appropriate
• Current referral setup (optional): Describe any existing program or past attempts so the skill can diagnose issues rather than starting from zero
• Tech stack (optional): Tools you use (Klaviyo, ReferralCandy, Smile.io, custom code) so recommendations are compatible with your infrastructure

Output format

The output is a structured referral program blueprint divided into six sections. First, a Program Summary with the reward model, referral flow diagram, and projected economics. Second, an Incentive Design section specifying exact reward types, amounts, and conditions for both referrer and referee, including tiered bonuses for power referrers. Third, a Sharing Mechanics section detailing channel-specific sharing flows for email, SMS, WhatsApp, Instagram, and unique referral links with sample copy for each. Fourth, a Fraud Prevention section listing specific rules — self-referral blocking, IP and device fingerprint checks, minimum purchase thresholds, velocity caps, and manual review triggers. Fifth, a Tracking and Attribution section covering UTM parameters, cookie windows, conversion pixels, and dashboard KPIs. Sixth, a Launch Checklist with phased rollout steps from soft launch to full promotion.

Scope

• Designed for: ecommerce operators, DTC brand teams, growth managers
• Platform context: Shopify, TikTok Shop, Amazon, WooCommerce, platform-agnostic
• Language: English

Limitations

• Does not integrate directly with referral software APIs — outputs a plan you implement in your chosen tool
• Reward economics are estimates based on inputs you provide; actual results depend on customer behavior and market conditions
• Does not provide legal advice on referral program compliance with local regulations; consult a lawyer for sweepstakes or cash-reward legality in your jurisdiction