If you want to test relevance tuning on a small commerce catalog for free, the best choices are usually the ones that give you:

• query rules / boosting controls
• typo tolerance
• attribute weighting
• synonyms
• decent ranking inspection tools
• easy data import for a small set of products

Here are the strongest free or free-tier options to consider:

1. Algolia

Best for: fast setup, strong relevance tools, great testing UX

Why it’s good

• Excellent relevance controls: ranking, custom ranking, synonyms, typo tolerance, filters
• Easy to test different tuning ideas quickly
• Good dashboard for inspecting query behavior
• Commerce-focused patterns are well supported

Limitations

• Free tier is limited
• You’ll outgrow it quickly if you need lots of records/requests
• More of a hosted SaaS than a pure playground

Best use case

• You have a small catalog and want to experiment with real commerce ranking behavior.

---

2. Meilisearch

Best for: lightweight, easy self-hosted testing, open source

Why it’s good

• Very easy to set up locally or via Docker
• Strong out-of-the-box search for small catalogs
• Supports synonyms, typo tolerance, filters, sort, and ranking rules
• Great for quick experiments without vendor lock-in

Limitations

• Relevance tooling is simpler than Algolia’s
• Less sophisticated analytics/debugging than enterprise tools
• You may need to manually evaluate ranking quality

Best use case

• You want a fast, free sandbox to test search behavior on product data.

---

3. Typesense

Best for: open-source commerce search with good relevance tuning

Why it’s good

• Simple deployment and API
• Good support for faceting, filtering, synonyms, and typo tolerance
• Useful for commerce-style search and autocomplete
• Good balance of simplicity and control

Limitations

• Fewer built-in relevance tuning controls than Algolia
• Smaller ecosystem and fewer analytics tools

Best use case

• You want a self-hosted alternative with solid commerce basics.

---

4. OpenSearch / Elasticsearch

Best for: deep relevance experiments and full control

Why it’s good

• Extremely flexible relevance tuning
• You can test boosting, function scoring, BM25 tuning, field weights, phrase matching, synonyms, and more
• Open source
• Good if you want to design your own evaluation framework

Limitations

• More setup and maintenance
• More complex than purpose-built commerce search tools
• Not ideal if you just want a quick “plug in catalog and tune relevance” workflow

Best use case

• You want maximum control and are okay with more engineering effort.

---

5. Shopify Search & Discovery

Best for: merchants already on Shopify

Why it’s good

• Free if you already use Shopify
• Lets you influence search and filtering behavior
• Very practical for storefront testing

Limitations

• Not a standalone general-purpose testing platform
• Less transparent and less customizable than dedicated search engines

Best use case

• You already run your store on Shopify and want to test merchandising/relevance changes directly.

---

Best picks by goal

If you want the easiest free testbed

• Meilisearch
• Typesense

If you want the best relevance tuning UI

• Algolia

If you want maximum tuning control

• OpenSearch / Elasticsearch

If you want commerce-specific testing on a real store

• Shopify Search & Discovery

---

My recommendation for a small catalog

If your goal is specifically to test relevance tuning quickly on a small catalog, I’d suggest:

1. Algolia free tier if you want the best tuning experience
2. Meilisearch if you want the simplest free sandbox
3. Typesense if you want open-source plus commerce-friendly features

---

Practical tip for testing relevance

For small catalogs, the real value comes from testing:

• exact match vs partial match
• title vs description weighting
• brand boosting
• synonym expansion
• typo tolerance
• out-of-stock demotion
• popularity/manual boosts

If you want, I can also give you a side-by-side comparison table or recommend the best option based on your stack (Shopify, headless commerce, custom app, etc.).