Magento quick start

Prerequisites

• Magento Open Source or Commerce 2.4.7+
• PHP 8.1, 8.2, or 8.3
• MySQL 8.0+ or MariaDB 10.4+
• Magento Marketplace credentials (public + private key) inauth.json
• Cron configured for the Magento default group

1. Install via Composer

Add our private Composer repository (one-time, you'll receive credentials when you sign up), then require the module:

composer config repositories.mba composer https://repo.marketbasketanalysis.com
composer require marketbasketanalysis/module-reports

2. Run setup

bin/magento module:enable MarketBasketAnalysis_Reports
bin/magento setup:upgrade
bin/magento setup:di:compile
bin/magento cache:flush

The newmba_reports_*tables are created via Magento declarative schema. Verify in the admin under Marketing → MBA Reports → Jobs.

3. Configure your LLM provider key

In the admin, go to Stores → Configuration → MarketBasketAnalysis → AI, pick a provider (Anthropic Claude or OpenAI), and paste your API key. This powers the AI-catalog engine, bundle-copy generation, and session-intent analysis. Skip this step if you only want the deterministic mining engines (FP-Growth, HUI, SQL pairs).

Bring-your-own-key: the module calls your chosen provider directly from your Magento install using the key you paste here. Your order data and prompts go to your own Anthropic or OpenAI account, not through our servers, and you're billed by the provider direct. Same architecture on the WooCommerce plugin.

4. Run your first mining job

From the command line:

bin/magento mba:run-job --engine=fp_growth --days=90

Or from the admin: MBA Reports → Jobs → New Job. The job runs synchronously for small catalogs (under 50k orders); larger jobs are queued and processed by cron.

5. Review your opportunities

Once the job finishes, browse MBA Reports → Opportunities. Each row is a ranked bundle candidate with confidence, lift, estimated profit, and a one-click Propose Bundle action. Approve a proposal and the module creates a Magento bundle product (in Draft) you can review before enabling.

6. Schedule recurring runs

In Stores → Configuration → MarketBasketAnalysis → Scheduling, set a cadence (daily, weekly, monthly) and a default window. The cron job mba_reports_run_next_queued_job processes the queue. You can also wire orders-velocity triggers to auto-mine when N new orders have landed since the last run.

Troubleshooting

• “Not registered handle” error , run rm -rf generated/code generated/metadata var/cache var/page_cache then bin/magento setup:di:compile.
• AI job returns 401, the API key in store config is empty or incorrect.
• Job stuck in “running” , check your cron logs and verify the default cron group is firing.

Next steps

• Wire the public API so external apps and AI agents can read your recommendations.
• Drop the MCP server into Claude Desktop or your agent host.
• Outgrown the shared tier? See Hosted Engine.