Shoppers search a catalogue in three registers: exact («WH-1000XM5»), sloppy («whireless heaphones», «yfeiybrb») and vague («gift for dad»). Most engines are good at one. Product search on AACSearch routes each to the retrieval that fits — exact identifiers stay exact, broken queries get rescued, vague ones match by meaning — and then lets the business decide what is shown first.
Install the 1C-Bitrix, PrestaShop or WooCommerce module, schedule a YML feed, or push products through the import API. Prices and stock stay in sync.
Choose searchable fields, facets and sort orders; filter out-of-stock items or push them down. Storefront widgets cover the box, results and filters.
Pin and boost with merchandising rules, send click and purchase events, and read which queries earn revenue and which end in nothing.
A query typed in the wrong keyboard layout, transliterated, or misspelled is still a purchase intent. When search answers it with an empty page, the shopper concludes the shop does not stock the item. The rescue ladder retries a zero-hit query through layout correction, transliteration, typo correction and broadening within the same request, and reports which strategy worked.
Whatever is still unanswered shows up in the zero-result report with the query text attached — a to-do list for synonyms, content and assortment.
The most relevant product may be out of stock, low-margin, or last season's. Merchandising rules put the business layer on top of relevance: pins, boosts, exclusions, banners and redirects, scheduled as campaigns. Sponsored placements stay separate from organic ranking, and every change can run as an A/B experiment before it becomes the default.
SKUs, article numbers and model codes match exactly and by prefix — they are not «corrected» into a different product.
Brand, price, size and attribute filters computed for the current query, on search and category pages alike.
Query suggestions come from real searches that return results, so the dropdown never leads to an empty page.
Search events tie queries to clicks, carts and orders — the funnel and revenue per query are reports, not spreadsheets.
{
"q": "yfeiybrb ,tcghjdjlyst",
"query_by": "name,brand,sku",
"facet_by": "brand,category,price",
"filter_by": "in_stock:true",
"sort_by": "_text_match:desc,popularity:desc",
"per_page": 24
}
// wrong keyboard layout -> rescued as «наушники беспроводные»
// -> hits, facet_counts, "query_rewrite": { "strategy": "layout" }Explore more
The best way to evaluate a search platform isn't reading one more comparison table.
Only then make the decision.