Shopify, PrestaShop, Magento: why their native search fails past 10,000 technical references?
The site has grown. The catalog now has tens of thousands of variants. And internal search, which worked well in the early days, has become slow or returns irrelevant results.
It's a familiar scenario: a B2B e-commerce site launches with a few thousand references, the platform's native search engine (Shopify, PrestaShop, Magento) does the job well. Then the catalog grows — new product lines, new suppliers, multiple variants — and search quality degrades gradually, without a clear reason why.
What's happening technically
Most e-commerce CMS native engines run on classic relational databases (SQL), which weren't originally designed for large-scale text search. This type of database handles transactions very well (orders, stock, payments), but noticeably slows down once the volume of text to index and the update frequency increase.
Past a certain reference count — the threshold varies by platform and catalog structure, but generally sits around a few tens of thousands of product sheets — two symptoms typically appear:
- Search response time degrades, noticeable to visitors
- Result relevance declines, since the native engine wasn't built to fine-tune ranking at high volume
The alternative: connect an external engine, without rebuilding the site
The most common reaction to this problem is assuming a platform migration is needed — a heavy, costly project, and often unjustified. In the vast majority of cases, the real fix is more targeted: connect a specialized external search engine (Typesense, Elasticsearch, Algolia, or a custom solution) that takes over search specifically, without touching the rest of the site's architecture.
The CMS keeps handling orders, stock, and payments as usual. Only search gets delegated to a tool built for indexing large text volumes, with latency that stays stable even past 100,000 references.
The right instinct isn't "we need to rebuild everything" — it's "we need to isolate the right component." And search is almost always isolatable without a rebuild.
How to know if this applies to you
If your catalog exceeds 10,000 references and you're seeing noticeable search slowdown, or relevance that's degraded as the catalog grew, that's the signal to watch for. An audit measures your current response time precisely and assesses whether an external engine is warranted — or whether simply retuning the native engine is still enough.