The Best Product Recommendation Engines
Every store wants the right product next to the one being viewed. This guide sorts the recommendation engines that do it — ready-made widgets, Shopify apps and developer APIs — by what each needs from you and where each one stops.
Where a recommendation engine gets its picks
Two kinds of signal, and every engine here mixes them. Which one it leans on decides how well it does on a new store, a new product, or a shopper on their first page.
Viewing: Studio Headphones
A sneaker added today, never viewed
Collaborative filtering
Recommendations from what other shoppers did. Algolia’s docs describe its Related items model as using “user interactions (collaborative filtering) and item attributes (content-based filtering)”, and publish what each model needs: 1,000 conversion events for Frequently bought together, 10,000 click and conversion events for Related items. Amazon Personalize’s User-Personalization-v2 needs at least 1,000 interactions.
Content-based and fallbacks
Recommendations from the product itself, which is how an engine copes with the cold start. Algolia’s image-based Looking similar “doesn’t require any events”. Personyze’s recommendations put fallbacks behind every widget — popular where the shopper is, then trending, then newest — so it is never empty on a new store.
How to choose a product recommendation engine
Most engines recommend well once they have data. Compare them on the things that differ:
Your platform
Any site through a tag, a plugin for your store, a Shopify-only app, or an API you build on.
Widgets or API
Ready-made, styled widgets a marketer places, or results your developers render.
What it needs to start
How many views and orders before a model works, and what it shows until then.
Email and beyond
Picks chosen when the email is opened, in your app, and in web push — from the same engine.
Control over the picks
Merchandising rules, filters, and putting what fits this shopper first.
How it is measured
Revenue per widget and per item, and a test against a group that sees no widget.
The best product recommendation engines
Grouped by how they reach your store, and ordered within each group by fit for a mid-size online store. Facts are from each vendor’s own pages in November 2026; prices change, so the tiers here are coarse.
Ready-made widgets for any major platform
A marketer sets them up and places them; the engine does the rest.
Nosto
Nosto’s Product Recommendations, part of its Product Experience Cloud, use “over 20 self-learning algorithms” — 1:1 personalized, cross-sell, visually similar, replenishment, geotargeted trending, and best sellers that need no personal data. Merchandisers can boost, bury, include or exclude products with no code, and email widgets pick products when the email is opened.
Strengths
- 20+ algorithms, with non-personal fallbacks
- Boost, bury, include and exclude rules
- Shopify checkout and thank-you page extensions
- Email picks at open time
Limitations
- No published prices: a base fee plus GMV and traffic
- No self-service trial; a proof of concept for qualified merchants
Personyze picks products for every page — home, category, product, cart, thank-you and search — each with its own algorithm, or Automatic, one widget that chooses per page, with fallbacks behind it. Similar Items (AI) finds a new product’s neighbours from titles, brands and categories, and Match the visitor puts the brand, size and price this shopper favours first. The same picks go into emails at open time, web push and your app; set up with AI, it places the widgets in your store’s look.
Strengths
- Any site, one tag; store APIs for Shopify, Magento, WooCommerce, BigCommerce, Shopware
- Preview as a visitor shows why each item is there
- Recommendations on the free plan
- Picks chosen when the email is opened
Limitations
- Open-time email recommendations start on the Growth plan
- Native iOS and Android SDKs are on Enterprise
Clerk.io
Clerk.io runs Search, Recommendations, Chat, Audience and Email on one store integration. Recommendations show complements, alternatives and next products from shopper and basket context, and Clerk sends its own personalized product emails as well as feeding Klaviyo, Mailchimp, Brevo and others.
Strengths
- Plugins for many platforms, plus an open API
- Native email recommendations
- No one-off implementation fee
- A free trial
Limitations
- Prices are not shown on its pricing page
- Recommendation emails count towards recommendation usage
Bloomreach
Bloomreach’s Product Recommendations sit in its Ecommerce Search, based on search history, browsing, purchases and product data: recently viewed, bestselling, frequently bought or viewed together, trending, in-stock, collections, and rule-based dynamic recommendations. Its email product inserts recommendations for each user.
Strengths
- Recommendations tied to on-site search
- A rule builder beside the automatic types
- Recommendations in its own email product
- 175+ integrations
Limitations
- No public rate card; from about $60K a year for smaller brands
- No free trial
Constructor
Constructor offers nine recommendation strategies, including complementary, alternative, bundles, abandoned in cart and query based, on the same engine as its search and browse. Placements are set up in its dashboard and served by API, with a Shopify app; email recommendations are picked at open time.
Strengths
- One engine with search and browse
- Email picks at open time
- A Shopify app and headless support
- Nine strategy types
Limitations
- No published pricing; demo only
- Integration is API- and developer-led
Apps for Shopify and similar stores
Installed from an app store; priced by your orders or revenue.
Rebuy
Rebuy is “purpose-built for Shopify”: recommendation widgets on any page, a Smart Cart with upsells, bundles, search and collections, and post-purchase offers. Recommendations from past orders can go into Klaviyo campaigns and flows.
Strengths
- The whole Shopify funnel, checkout included
- Bundles and a bundle builder
- Klaviyo email recommendations
- From $25 a month, with a 14-day trial
Limitations
- Shopify only
- Price depends on monthly Shopify orders
LimeSpot
LimeSpot does product recommendations, upsells, bundles and personalized discounts on website, mobile and email, on Shopify, BigCommerce and SHOPLINE. Its Klaviyo integration puts recommendation widgets into emails.
Strengths
- All features in one plan
- 15-day free trial, no setup fee
- Email recommendations through Klaviyo
- Bundles and personalized discounts
Limitations
- Only Shopify, BigCommerce and SHOPLINE
- From $150 a month, tied to store revenue
APIs for developers
You send the events and draw the results yourself.
Algolia Recommend
Five models — frequently bought together, related items, trending items, trending facet values and the image-based looking similar — served by API and by ready UI components for JavaScript and React. Models retrain once a day.
Strengths
- Free tier of 5K requests a month
- Image-based model needs no events
- UI widgets for JS and React
- Published data minimums per model
Limitations
- Most models need event data first
- Widgets are code a developer adds
Recombee
A recommendation and search engine behind a REST API with SDKs in many languages, plus a no-code widget. Each scenario is one place recommendations appear — site, app or email — and a rules language sets filters and boosters.
Strengths
- A free plan with published limits
- Published paid prices
- Its own rules language
- Web, app and email
Limitations
- Expect API work; no store plugins on the pages read
- Paid tiers rise steeply
Amazon Personalize
A fully managed AWS service that generates item recommendations from your data, with ready ecommerce recommenders such as “Recommended for you” and “Frequently bought together”, real-time API calls, and batch jobs for email lists.
Strengths
- Pay for use, no upfront commitment
- Real-time and batch
- Ready recommenders or your own recipes
- Free tier for the first two months
Limitations
- API only: you build the display
- Real-time inference has a minimum charge
What you get inside Personyze
Product recommendations
An algorithm for every page, fallbacks behind it, and quick add to cart from the widget.
See recommendations →Set up with AI
Connect your store; it places widgets on home, product and cart pages in your look, unpublished until you say.
Read the guide →Email recommendations
Products picked for each recipient when they open the email, in the email tool you already use.
Read the guide →Product recommendation engines compared
Where each one runs, who builds the widget, and what it costs to start.
| Engine | Platforms | Widgets or API | Email picks | Free plan or trial | Pricing tier |
|---|---|---|---|---|---|
| Nosto | Major platforms, plus APIs | Ready-made | Yes, at open | Proof of concept | Contact sales |
| Personyze | Any site; 5 store APIs | Ready-made, plus JSON | Yes, at open (Growth+) | Free plan; 14-day trial | Free–$$ |
| Clerk.io | Many plugins, plus API | Ready-made, plus API | Yes, native | Free trial | Usage-based |
| Bloomreach | Any site, 175+ integrations | With its search | Yes, its own email | No | $$$ |
| Constructor | Any site by API; Shopify app | API; Shopify components | Yes, at open | Demo | Contact sales |
| Rebuy | Shopify only | Ready-made | Through Klaviyo | 14-day trial | $ |
| LimeSpot | Shopify, BigCommerce, SHOPLINE | Ready-made | Through Klaviyo | 15-day trial | $$ |
| Algolia Recommend | Any site by API | API, plus UI components | Not stated | Free plan | Free, then usage |
| Recombee | Any site or app by API | API, plus a widget | Yes | Free plan | Free–$$$ |
| Amazon Personalize | AWS API | API only | By batch job | Two-month free tier | Pay as you go |
Frequently asked questions
What is a product recommendation engine?
What is the difference between collaborative and content-based filtering?
How much data does a recommendation engine need?
Can product recommendations go into emails?
Do I need a Shopify app or a platform?
How do you measure product recommendations?
Put the right product next to every product
Connect your store, let the AI place the widgets in your look, and check them as a real shopper before anything goes live.