The Related Products block is empty on most of your product pages

Adobe Commerce and Magento Open Source show Related Products, Up-Sells and Cross-Sells only for the links someone set, product by product. Nobody has time to do that for two thousand products, so on most pages the blocks are empty – and on the pages that were curated once, they point at last season. The theme has a place for recommendations; the catalogue does not fill it.

What it is for

Smart Product Recommendations Engine fills those same three blocks from a few rules over your product attributes, configured once per position instead of once per product. "Same colour", "same manufacturer", "overlapping words in the name", "a price within 20 %" – each is a rule, and the rule compares the product being viewed with the rest of the catalogue.

A fashion store sets one same-colour rule for Related Products, and every product page shows colour-matched items. It sets a same-brand rule with a higher price band for Up-Sells, and a tag rule for Cross-Sells, which the cart page evaluates for every line in the cart and never suggests an item that is already there.

The engine runs on your server and reads your live catalogue. It adds no database table, no index and no cron job. It is deterministic: no model and no shopper data – the same rules over the same catalogue give the same pool of products.

What changes in the merchant's week is that the three blocks stop being a maintenance chore and become a setting.

How it can be used

A catalogue whose attributes already describe it. Colour, brand, material, a tag attribute like summer/sport/casual, or simply the product name. Seven match types cover them: token overlap, slash-tag overlap, same value, any value matches, price range, starts with and exact match. Rules run in priority order, a product never appears twice, and a position never exceeds its Max products. The trade-off is plain: thin attributes make thin matches, and enriching the catalogue comes first.

A store that has curated some links and wants to keep them. Each position has four modes. Append after native – the default – shows your hand-set products first and adds rule matches after them; Fill-up tops up to the maximum; Replace native and Auto only hand the block to the rules. Curation stays where it is worth the time.

Luma and Hyvä side by side. The engine plugs into the platform's own Related, Up-Sell and Cross-Sell blocks, and into Hyvä's product-list view model, so the theme's markup and styling carry the result. One package and one configuration serve both storefronts.

Different logic per store view. Every setting, rules included, is store-scoped. A master switch returns the storefront to native behaviour without losing a rule, and the extension ships switched off.

How it compares

Measured on 2026-09-17: fourteen Marketplace listings opened in a browser, the price read off each edition.

The core of the category is eight "automatic related products" listings, and every one of them uses rule or attribute conditions. Their Magento Open Source prices: Sparsh 0.00; Milople 125.00, Mageside 159.00 and MageAnts 169.00 one time; MageComp 79.00 a year then 40.00, Mirasvit 219.00 then 132.00, Aheadworks 254.00 a year, Amasty 309.00 then 185.00. Over two years the last three cost between 351 and 508. Around them sit recommendation engines with another premise: PPL builds a co-purchase index from order history, PIT sends each visitor's behaviour to OpenAI, and Adobe's own Product Recommendations is a machine-learning service for the Adobe Commerce editions only.

What favours this extension, each point demonstrable:

• A one-time USD 99.99, below every paid one-time peer and under a third of what Mirasvit, Aheadworks or Amasty cost over two years.
• It fills the blocks the store already has rather than adding blocks of its own, so there is no second layout to style and nothing to position.
• Nothing collected, nothing to rebuild. No visitor tracking, no index, no cron, no table – so no stale statistics and no personal data.
• Luma and Hyvä from one package, where MageComp sells a Hyvä add-on and PPL a separate Hyvä package priced higher.

Where a competitor is the better buy, it is worth naming. If you want "customers who viewed or bought this also…", Amasty, Mirasvit, Aheadworks, Mageside and Milople read your store's behaviour, and this extension never does. If you need to measure the blocks, Mirasvit reports impressions, clicks, orders and revenue; Amasty, Aheadworks and Milople report too. If recommendations belong elsewhere – home page, category page, account page, a widget – or a rule should apply only to one customer group or date range, the peers add those positions and conditions. If your storefront is headless, MageComp and Mirasvit state GraphQL support. And if the budget is zero, Sparsh's Auto Related Products is free, rule-based, and rated 4.25 from four reviews.

What it does not try to be

It does not learn and does not personalise: every visitor to a store view sees the result of the same rules. It adds no blocks, widgets or positions, and has no bundle with a combined add-to-cart. A rule applies to the whole store view; there is no per-category condition. It has no reports, A/B testing, console command or cron job, and does not change the GraphQL API. It calls nothing and sends nothing. The rest is in the FAQ on the extension's page.