E-commerce · Catalog search

Catalog search for E-commerce

For merchants, catalog search on KERNSTACK is merchandising infrastructure you own: a text index declared over the live products table, facet and price checks against the rows themselves, a synonym table the merchandising team edits as data, and a zero-results log that writes next week’s priorities. The index cannot drift from the catalogue because it is the catalogue, viewed through a declaration.

Full-Text Search NoSQL Tables Functions (WASM)CDN Edge

Talk to us Read Technical Documentation → Software Lifecycle Cloud →

Domain Architecture for E-commerce

The domain work is vocabulary and truth. Products carry the fields shoppers think in — title, category, attributes, the words support keeps hearing — and the index covers exactly those. Synonym rows (sofa/couch, trainer/sneaker, the brand nickname only your customers use) accumulate as a business asset beside the products, and campaign boosts are declared rules evaluated against live price and stock, so a promoted line is promoted at its true price while it is actually available.

Compliance, Data Residency & Governance

Price-display honesty is the quiet regulatory win: because search results, category pages, and checkout read the same product rows, the price a shopper finds is the price they pay — the inconsistency class that generates consumer complaints and regulator letters is removed architecturally. Catalogue and query data remain in the European zone, and zero-result logs are ordinary project data under the project’s retention rules.

Commerce regulation is mostly about promises and records: the price displayed is the price charged, orders and invoices are retained and retrievable, refunds follow a trail, and customer data obeys the usual European duties of residency, erasure, and evidence. KERNSTACK’s shape supports those duties plainly — catalogue prices read live from one table (so search results cannot quote a stale price), order states derive from stored payment events, retention is declared per record, and the whole shop’s data sits in one European zone a merchant can name in their privacy documentation.

Operational Impact on the Software Lifecycle Cloud

Merchandisers stop filing tickets. Synonym fixes land the day the pattern is spotted; seasonal boosts are reviewed changes with dates, so January can ask what December’s ranking did and get an answer; and the weekly zero-result list replaces opinion with customer vocabulary in the roadmap meeting. Search quality becomes a compounding, owned asset instead of a vendor’s black box.

Key Domain Entities & Workflows

Primary E-commerce Data Entities: products, carts, orders, payment events, fulfilment tasks
Governed Lifecycle Workflows: catalogue publish, checkout, refund approval, fulfilment exception, returns intake
Integrated Platform Services: search, tables, functions, cdn

Compare this integrated blueprint with the general Catalog search reference architecture or explore our European Sovereign Cloud guarantees.

E-commerce · Catalog search Flow

IngressShopper query at search routecatalog-search

StateText index over live productsproject resource

ReviewFacets checked against rowsdeclared in stack.json

stack.json — E-commerce Catalog search plan · deploy

{
  "resources": {
    "functions": [{ "name": "search" }],
    "tables": [
      { "name": "products", "key_name": "product_id", "key_type": "S", "indexes": [] },
      { "name": "synonyms", "key_name": "term", "key_type": "S", "indexes": [] },
      { "name": "zero-results", "key_name": "query", "key_type": "S", "indexes": [] }
    ],
    "search_indexes": [{ "name": "catalog", "index_type": "text",
      "sources": [{ "type": "tables", "name": "products", "fields": ["title", "description", "category", "tags"] }] }]
  }
}

Estimated Footprint · Team Plan ~1.500 credits / mo

Assumes a 20k-product catalogue and half a million searches a month, mostly index and keyed reads inside function compute.

Metered from one shared balance across compute (0.01 cr/GiB-s), storage (0.04 cr/GB-hr), network (9.00 cr/GiB), and catalog models. See full credit reference →

FAQ

Frequently asked questions

Who can change synonyms and boosts?

Merchandisers, through data and declared rules — no deploy, no search-vendor console. Changes carry authors and dates, so a ranking experiment can be attributed and reverted like any other business decision.

How do zero-result queries reach the team?

They are rows. A weekly look at the zero-results table, ranked by frequency, is the merchandising agenda: add the synonym, stock the product, or fix the category name — written in the customers’ own words.

Does search stay correct during a sale?

Yes, structurally: results price and filter against live rows, so a price drop or a sell-out is reflected the moment it is true. Sale-morning catalogue edits do not wait for a sync window while traffic arrives.

Connected in the Project Graph

Related capabilities, architectures, and comparisons

Ship your stack on the Software Lifecycle Cloud.

Closed preview on dedicated European infrastructure. Bring a repository or start from a blueprint.

Talk to us See pricing & credits →