MBA
Profit-aware bundle ranking with explainable math

More profit per bundle,
not just more attach.

We rank bundles by margin-weighted profit, not raw attach rate, and show the support, confidence, and lift behind every one. So you can audit each recommendation back to the orders it came from. The whole suite costs less than one single-purpose widget, flat and predictable, with no GMV surprises. Also agent-ready: the same intelligence is callable over MCP. No black box. No vendor magic.

First-party by design. Your data stays in your store. No pooled training models.

Free $0 · Plus $99/mo flat · Flat, never metered on GMV · BYO LLM key

Pizza ovenPizza peel
83% confidence4.20× lift1.2% support
Pizza ovenInfrared thermometer
41% confidence2.80× lift0.8% support
Pizza peelPizza stone
67% confidence3.10× lift1.5% support
The moment

Agentic commerce just became the operating model

At Adobe Summit 2026, Adobe blessed open MCP as the standard for the agentic-commerce wave, partnering with Anthropic, OpenAI, Google, and Microsoft. AI traffic to US retail grew +269% YoY in March 2026. The platform layer is being agent-ified. The question becomes: what does the agent recommend?

Platform layer

Adobe Commerce MCP, Salesforce, others

Exposes catalog, cart, pricing, inventory, checkout to agents. Tells the agent what's available.

Intelligence layer (us)

MarketBasketAnalysis MCP

Exposes co-purchase intelligence as agent-callable tools. Tells the agent what to suggest: recommendations, kit completion, and basket cohesion.

Reasoning layer

Claude, GPT, Cursor, Cline, your agents

Composes both layers to drive the workflow: shopping assistant, merchandiser copilot, restock planner.

AI agent traffic

+269%

YoY to US retail, March 2026

MCP tools we ship

18

Agent-callable Basket AI tools (18 off Shopify), any MCP host

Platforms supported

5

Shopify, BigCommerce, WooCommerce, Magento, OroCommerce, same intelligence

Not another upsell widget.

This is a bundle and kit creation engine, powered by your real transactions.

Typical Upsell Apps

  • Frontend-only
  • Black box
  • Optimizes clicks

Manual Merchandising

  • Time-consuming
  • Guesswork
  • Hard to measure

Market Basket Analysis

  • Bundles, cart drawer, post-purchase
  • Profit-ranked + A/B tested for lift
  • Explainable + reports
Shipping today

Three surfaces. One intelligence layer.

One mining engine feeds the admin actions, the storefront widgets, AND the agent / MCP API.

AdminShopify + BigCommerce + WooCommerce + Magento + OroCommerce

Admin intelligence + bundle builder

  • Mining jobs (manual + scheduled + orders-velocity triggered)
  • Rules + opportunities grids with filters and live status
  • Bundle proposals turn into real catalog bundles with components
  • AI-generated bundle copy + multi-currency performance dashboard
  • Weekly digest emails + operator overrides per opportunity
Explore admin
StorefrontWooCommerce

Storefront recommendation widgets

  • "Frequently Bought Together" widget on the WooCommerce product page
  • Shortcode, Gutenberg block, and theme template tag, all from one template
  • Powered by precomputed rules. Sub-millisecond fetch
  • Hides itself when no rules exist for the source product
Preview widgets
Agent / MCPAny MCP-compatible host

Agent skills + API access

  • Basket AI: 19 agent-callable MCP tools (18 on non-Shopify platforms), any MCP host
  • Tools: get_recommendations, find_substitutes, get_bundle_for_cart, score_cross_sell, analyze_basket, predict_reorder, propose_subscription_bundle
  • Public REST API with mintable per-shop API keys
  • MCP server (npm: @marketbasketanalysis/mcp)
  • One-line setup for Claude Desktop / Claude Code / OpenAI Agent SDK / Cursor / Cline
See the API
Shipping today

Conversion surfaces, on all five platforms

The recommendation engine now drives the moments that actually move revenue: the cart, the checkout, the experiment, the bundle. Every one ships on Shopify, BigCommerce, WooCommerce, Magento, and OroCommerce, not just one storefront.

Cart drawer with AOV progress

An opt-in slide-out cart with a free-shipping / AOV progress bar and one to three in-cart recommendation upsells. The upsells recompute as the cart changes, so the next-best add-on is always live.

Merchant A/B testing

Run experiments on the recommendation surfaces with a deterministic control / treatment split. The results dashboard reports impressions, conversions, conversion rate, AOV delta, and lift with a 95% confidence interval via a 2-proportion z-test. The Frequently Bought Together widget is wired end to end.

Post-purchase upsell

After checkout, the thank-you / order-confirmation surface shows ranked recommended add-ons with one-click add. Ungated on the entry tier, on the same co-purchase data as the rest of the engine.

Broadened bundle formats

Beyond real-SKU bundles, pick a bundle format: mix-and-match / build-a-box (BYOB), volume / quantity breaks (tiered), BOGO / Buy X Get Y, and a virtual-bundle render mode that groups existing products with no duplicate SKU.

One engine. Five platforms. No single-storefront lock-in.See how the conversion surfaces work
Consulting

We can implement this for you in 2-4 weeks.

Get bundles live, priced correctly, and measured, without adding workload to your team.

QuickStart

1–2 weeks

Install, configure, first analysis run, and top opportunities report.

Learn more

Bundle Launch Sprint

2–4 weeks

Create optimized bundles, set pricing strategies, and establish a measurement plan.

Learn more

Ongoing Optimization

Monthly

Re-mining, seasonal reviews, bundle backlog management, and performance reporting.

Learn more
The admin

Admin intelligence that turns data into bundles

A complete admin surface, shipping today on every supported platform, for discovering co-purchased patterns, scoring opportunities by profit potential, and creating catalog-ready bundles.

marketbasketanalysis.com
Schedule analysis runs with guardrails and verbose logs
marketbasketanalysis.com
Raw association rules for transparency and auditability
marketbasketanalysis.com
Ranked opportunities with estimated revenue and profit lift
marketbasketanalysis.com
Bundle proposals with pricing strategies
marketbasketanalysis.com
Create draft bundles and manage lifecycle
marketbasketanalysis.com
Track bundle conversion, revenue, and profit
marketbasketanalysis.com
Shareable executive summaries
marketbasketanalysis.com
Control thresholds, engines, and safety limits
marketbasketanalysis.com
Built-in help for teams new to MBA
Agent surface

Basket AI: 19 agent-callable MCP tools. Any MCP host. One npm install.

Drop our MCP server into the agent host of your choice. One snippet, one restart, 19 agent-callable tools (18 on non-Shopify platforms, where predict_reorder is Shopify-only) the agent can call against the merchant's real co-purchase data. The agent commerce window opened in March 2026. This is our part of it.

{
  "mcpServers": {
    "marketbasketanalysis": {
      "command": "npx",
      "args": ["-y", "@marketbasketanalysis/mcp"],
      "env": { "MBA_API_KEY": "mba_live_..." }
    }
  }
}
  • ShopFind what goes with a product
  • ShopReplace an out-of-stock SKU
  • ShopComplete the kit for this cart
  • ShopValidate a pair before suggesting it
  • ShopCheck whether a bundle holds together
  • ShopPredict next-purchase dates per product
  • ShopPropose a recurring subscription kit
Compatible hosts
Claude DesktopClaude CodeOpenAI Agent SDKCursorClineany MCP stdio host
Why MBA

More than an upsell widget

Purpose-built for operators who need transparency, control, and profit-aware automation.

Creates Bundles & Kits

Goes beyond simple recommendations. Creates actual catalog products with SKUs, pricing strategies, and inventory settings.

Profit-Aware Scoring

Opportunities ranked by estimated profit lift with margin floor protection. Never sacrifice margin for volume.

Explainable Metrics

Every recommendation backed by support, confidence, and lift scores. Audit-friendly logs and transparent decision-making.

Time Windows & Trends

Compare 30-day vs 90-day windows. Spot emerging opportunities and declining patterns before they impact revenue.

Admin-Native, Privacy-First

Built into the admin you already use across Shopify, BigCommerce, WooCommerce, Magento, and OroCommerce. Local data processing by default. AI features are bring-your-own-key: Anthropic or OpenAI, billed direct to your provider account.

Weekly Action Plan

Every mining run produces a ranked weekly plan: the highest-profit bundles and substitutions to ship next, with bulk Opportunities actions to push them live in one click.

Conversion Surfaces, Five Platforms

The same engine drives an opt-in cart drawer with an AOV progress bar, a post-purchase upsell on the order-confirmation page, and broadened bundle formats (BYOB, volume breaks, BOGO, virtual bundles). All on Shopify, BigCommerce, WooCommerce, Magento, and OroCommerce, not a single storefront.

Measurable Lift, Built In

Merchant-run A/B tests on the recommendation surfaces split traffic deterministically and report impressions, conversions, conversion rate, AOV delta, and lift with a 95% confidence interval via a 2-proportion z-test. Prove the revenue, do not guess at it.

FAQ

Frequently asked questions

Everything you need to know about market basket analysis and how this module works.

Ready to turn your order data into revenue?

Install on your platform in under 10 minutes. Or book a consulting call and we'll do the launch for you.