Search Ranking for eCommerce: When Algolia Is Overkill (and What to Use Instead)
Algolia is a fantastic product. It's also priced for scale, and for eCommerce sites below a certain volume, you can get 80% of the experience for 5% of the cost. Here's an honest map.
Key takeaways
- Algolia wins for: high-volume sites, complex relevance tuning, instant search UX.
- Algolia is overkill for: small-mid catalogs, basic search needs, budget-constrained teams.
- Alternatives: Meilisearch (self-hosted, free), Typesense (self-hosted, free), Postgres full-text + pg_trgm (already in your stack), ElasticSearch (heavy).
- Real switching cost: 1-2 weeks of engineering.
When Algolia wins
- 1M+ products in catalog
- Sub-100ms latency required globally
- Complex personalization + ML ranking
- Marketing team needs no-code relevance tuning
- High-volume search (1M+ queries/month)
When Algolia is overkill
- <100K products
- Basic search needs (autocomplete, fuzzy match, basic filters)
- Tight budget
- Team capable of self-hosting
Alternatives
Meilisearch
Self-hosted, Rust-based, designed for "good defaults." Instant search out of box. 80% of Algolia DX at 5% of cost (you pay for hosting only).
Typesense
Similar story. Open-source, self-hosted or managed. Fast, good defaults.
Postgres full-text + pg_trgm
If you're already on Postgres (most eCommerce teams are), the built-in full-text + trigram extensions handle basic search remarkably well. Free.
ElasticSearch
Powerful but operationally heavy. Justify it with scale.
What switching looks like
Indexing: 1-2 days. Reproducing relevance rules: 3-5 days. Frontend integration: 2-3 days. UAT: 1 week. Total: 2-3 weeks.
The hard part is recreating the relevance tuning a marketing team has built up in Algolia. Document it before you migrate.
What we recommend
Below ₹20 crore revenue with <100K products: Postgres full-text or Meilisearch.
₹20-100 crore revenue with growing catalog and need for marketing-team-driven relevance: stay on Algolia or evaluate Typesense for cost savings.
₹100+ crore with complex personalization needs: Algolia or build custom on Elasticsearch.
FAQs
Can Postgres really do eCommerce search? Yes for catalogs up to ~50K SKUs with reasonable query patterns.
Algolia vs Typesense managed? Typesense Cloud is cheaper; Algolia has more polish.
Recommendations vs search? Separate problems. Don't conflate.