Upsell Mapper

Map product relationships and purchase sequence patterns to identify the best cross-sell triggers at checkout, post-purchase, and in creator content.

Security Scan

The skill's claims, required inputs, and runtime instructions are internally consistent: it works from pasted CSV/order data, requests no credentials or installs, and does not try to access unrelated system resources.

Purpose & Capability

Name/description align with the requested inputs and outputs. All declared inputs (SKU list, order history, placements) are appropriate and sufficient for the stated analytics task. There are no unexpected environment variables, binaries, or config paths required.

Instruction Scope

SKILL.md confines runtime behavior to ingesting user-supplied CSV/table data and producing an affinity map and playbooks. It explicitly states it does not connect to live ecommerce APIs and does not instruct reading system files or other environment state.

Install Mechanism

No install spec and no code files — instruction-only skill. This is the lowest-risk distribution model and matches the described manual data-paste workflow.

Credentials

The skill requests no credentials or secrets, which is proportionate. Note: the required inputs are order histories and notes that may contain customer PII; while not a mismatch, this is a privacy consideration for the user to manage.

Persistence & Privilege

always:false and the skill does not request persistent system presence or modify other skills. Model invocation is allowed (platform default) and is not otherwise elevated.

Assessment

This skill is coherent and runs from data you paste in, but before using it: (1) strip or anonymize customer PII (names, emails, full addresses, payment tokens) from exports you paste; (2) share only the columns needed (SKU, order date, qty, price, return flag) to reduce exposure; (3) verify the recommendations against merchandising judgment and seasonality before deploying; and (4) if you prefer tighter control, run the analysis locally or in a trusted environment where pasted data never leaves your systems.

Most ecommerce brands sprinkle cross-sell widgets across the funnel without evidence that the pairings actually convert. This skill turns raw order data and content placements into a concrete upsell map that tells you which SKU should be offered after which, at which funnel stage, with which angle, so every cross-sell slot on your storefront and creator content has a tested reason to exist.

Use when

• A Shopify, TikTok Shop, or Amazon seller asks "what should I put in my post-purchase upsell slot?" or "which products sell best together after X bestseller?"
• A brand team wants to restructure the checkout add-on widget, bundle suggestions, or "frequently bought together" module on product detail pages
• A creator or affiliate team needs a shot list of companion products to weave into TikTok Live, UGC scripts, or email flows after a hero SKU sells
• An operator is planning a loyalty or replenishment program and needs a logical product adjacency graph to build the reorder sequence

What this skill does

Takes a list of SKUs, order line-item data, and optional review or return notes, then builds a directional affinity graph: for each source product, it ranks the top candidate follow-up products by attach rate, gross margin contribution, average time between purchases, and return correlation. It then groups these into three placement-specific playbooks — checkout add-ons, post-purchase one-click offers, and creator content pairings — because each slot rewards a different kind of upsell (impulse, convenience, demonstration).

Inputs required

• SKU list with categories, prices, margins (required): CSV or pasted table covering at least the top 50 active SKUs
• Order history sample (required): order ID, line items, quantities, purchase date — 3 to 12 months is ideal
• Current upsell placements (required): which slots exist today (PDP, cart, checkout, post-purchase, email, creator content)
• Return or refund notes (optional): helps downweight high-return pairings
• Creator content inventory (optional): so pairings can be mapped to specific video formats or hooks

Output format

A three-part deliverable. First, the affinity matrix: a ranked table of source SKU, recommended follow-up SKU, attach rate, incremental margin per thousand orders, and confidence note. Second, placement-specific playbooks: for checkout, post-purchase, and creator content, a shortlist of the top five pairings with the recommended copy angle for each slot. Third, a prioritized rollout plan: which pairings to test first based on traffic volume, implementation effort, and expected margin uplift over a four-week test window.

Scope

• Designed for: ecommerce operators, DTC brands, TikTok Shop sellers, and marketplace sellers with at least a few months of order history
• Platform context: platform-agnostic, with specific placement notes for Shopify, Amazon, and TikTok Shop
• Language: English

Limitations

• Does not connect to live ecommerce APIs; works from exported data you paste in
• Statistical confidence suffers with fewer than ~500 orders per source SKU; the output will flag low-data pairings
• Does not replace merchandising judgment for brand-sensitive pairings or seasonal considerations