From real order history
The plugin analyses which products are actually bought together - with a minimum co-purchase count and confidence per product pair. Statistics instead of gut feeling.
AI Cross-Selling.
AI-driven cross-selling assignment from real order history and catalog - market-basket analysis plus AI scoring, straight into native Shopware cross-selling.
What the plugin does
Cross-selling based on data.
Instead of maintaining cross-sells manually or guessing, the plugin combines statistical analysis of real order data with AI scoring - and writes only reviewed, sensible assignments to native Shopware cross-selling.
Complement, not coincidence
An AI rates every suggested pair: is B a real accessory/complement to A - or just a substitute (competing product)? Only sensible combinations are kept.
Even without order history
New products are not left out: the plugin records which products are viewed together within a visit and builds cross-selling from that. Only product pairs with a counter are stored, no personal data.
Understands content, not just numbers
Using AI embeddings, the plugin finds products that are related in substance - based on name, description and manufacturer. It works without purchase history, shared categories or maintained properties, which makes it the rescue for thin data.
Several blocks at once
Each profile is its own preset with its own sources, thresholds and automation - and produces its own cross-selling block on the product page. "Frequently bought together" and "Matching accessories" run in parallel, each controlled separately.
Traceable per pair
For each suggestion the AI provides a short reason why the combination fits - transparently visible in the review workflow.
Accept or reject
Suggestions land in a pool for review. Accepted and rejected assignments are preserved and never overwritten on the next analysis.
No custom frontend
Reviewed suggestions are written to native Shopware cross-selling - no extra storefront element, full theme compatibility.
Why it works
Data plus judgement.
The combination of statistical analysis and AI scoring delivers better cross-sells than either method alone.
Why AI plus statistics?
Numbers show the what, the AI the why.
Pure market-basket analysis finds pairs that are often bought together - but cannot tell whether that makes sense (printer + cartridge = good, two competing printers = bad). The AI layer filters exactly that out and adds cold-start cases for which there is no sales data yet.
Fitting add-on purchases
Real complements increase average order value without annoying the customer with mismatched suggestions.
Automatic suggestions
Instead of setting thousands of cross-sells by hand, you only review the suggested pairs - a fraction of the effort.
Complement vs. substitute
The AI separates real accessories from competing products - your cross-sells stay relevant instead of redundant.
Configuration
Fully in the Shopware admin.
From provider choice via thresholds to the stats/AI blend - all configurable in the admin. No code changes required.
Freely selectable
OpenAI, Anthropic, Mistral or Gemini - enter an API key, pick a model, test the connection. You use your own account; nothing is routed through us.
Keep the strongest N
Per source product, capped to the strongest N suggestions - no endless lists, only the best matches.
Combinable & optional
Seven sources can be combined: joint purchases, products viewed together, categories, properties, manufacturer, category bestsellers and semantic similarity via AI embeddings. AI scoring remains optional.
Automatic analysis
The analysis runs as a scheduled task in the background and rebuilds the suggestion pool regularly.
Min. co-purchases & confidence
Minimum values for co-purchases and confidence filter statistical noise before the AI even scores.
Trigger manually
The analysis can also be triggered manually via CLI - handy for initial seeding or tests.
Control
Fine control per profile.
Not every block should behave the same way. That is why each profile gets its own rules - from visibility to automatic acceptance.
Per shop
A profile can be bound to a sales channel. Only products visible in that shop are considered, and the suggestions belong to that channel - cleanly separated across multiple shops.
Leave brands out
Per profile you can exclude manufacturers whose products should never be suggested - discontinued brands or suppliers with poor margins, for example.
No empty blocks
If the sources return too few matches, the plugin tops up with category bestsellers until the requested minimum is reached. Every product page shows something useful instead of an empty block.
Automatic above a score
Configurable per profile: from which score suggestions are accepted without asking - optionally only genuine AI complements. One profile rolls out on its own, another stays under manual review.
Complement, not competition
The AI tells complements from alternatives. One checkbox keeps only true complements in the suggestions - no competing products on your own product page.
After profile changes
If a profile was renamed or deleted, a cleanup function removes the orphaned open suggestions. Pairs already accepted or rejected stay untouched.
Technical details
Clean Shopware integration.
Market-basket analysis, AI scoring, review pool and writing to native cross-selling - as a scheduled task, without a core patch.
Compatibility
Market-basket analysis
The MarketBasketAnalyzer builds ordered product pairs (A → B) with support, confidence and lift from order history. Co-occurrence is deduplicated (a < b), both directions are scored.
AI scoring
The AiScoringService rates each pair (complement vs. substitute, cold-start) and provides a reasoning text. Statistical and AI signal are blended.
Review pool
Suggestions land in a pending pool. Accepted/rejected assignments are never overwritten - repeated analyses are safe.
Writing to cross-selling
The CrossSellingWriter writes reviewed suggestions to native Shopware cross-selling. No custom storefront element needed.
Scheduled task
The CrossSellingAnalysisTask rebuilds the suggestion pool periodically - runs with the normal scheduled-task worker.
Availability
Available in the official Shopware Store - with auto-update notifications in the admin.
Availability
Through the store or directly from us.
Two routes, the same feature set - pick whichever suits your setup.
Available right now
Monthly9,99 €
Half-yearly49,99 €17 % cheaper
Best valueYearly89,99 €25 % cheaper
All prices net, plus VAT.
You buy the licence directly from us and get an invoice from us. Install by upload in the admin without SSH, updates through our licensing platform, support straight from the developer.
In preparation
The plugin will be listed in the official Shopware Community Store. Installation and automatic updates then run through the plugin manager, billing through your Shopware account. We will link it here as soon as it is live.