E-commerce terms, from faceted search to checkout

Product discovery in e-commerce is everything that helps a customer get from "I need something like this" to the right product: search, filters, sorting, product relationships, comparison and guided selection. In product management the same words mean deciding what to build; in a store they describe the customer's path through the catalog. It works only when product data, catalog structure, search, filter logic and interface are designed together, and many problems are solved with clean data and clear rules rather than AI. Baymard Institute's 2026 benchmark rated search on 56% of the sites it studied as mediocre or worse.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Site search is the search box of an online store and the engine behind it, the way customers find what the menu does not show. A search that matches only the words of the product sheet fails real queries: the sheet says "duvet cover 200x220, satin, 100% cotton", while the customer types "soft bedding for a double bed". Its queries are also free research into what people want and what they call it. On a very large assortment, search-first navigation, with a full-width search bar as the main tool, can be the design itself.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Faceted navigation is the set of filters in an online store, such as brand, size, color and price, that narrow down a category. Every combination can create its own URL, so a catalog can generate thousands of near-identical pages that waste crawling and compete with the main categories. The fix is a rule in the code: index only the combinations with real search demand and keep the rest out of the index. A shoe category with 10 colors, 12 sizes, 30 brands and 5 price ranges alone can produce 18,000 filter combinations.

More: Filters and Faceted Navigation: How Online Stores Create Thousands of Duplicate Pages

Faceted search is narrowing search results or a category by several independent attributes at once, such as size, material, brand and price, each of which is a facet. In a well-built store it shows only the facets relevant to the category (bed linen needs size and fabric, not the voltage of lamps), hides combinations that cannot exist and shows live counts, so no click leads to an empty page. Values combine with AND between facets and OR inside one facet. When filtering slows down as the catalog grows, the fix is usually indexing and query structure, not a new platform.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A facet is one dimension of filtering in a store, built on a product attribute with a defined list of values: size, color, material, brand or price range. Facets combine with AND between them and OR inside one, and a product with several variants matches when any single variant fits every selected facet. A facet is only as good as its data: in a pearl jewelry catalog, size in millimeters became the main facet, in 0.5 mm steps up to 9 mm and 1 mm steps above, and it could not launch until every variant had a minimum and maximum size.

More: The Website Design Process Most Agencies Show You Is Wrong

Facet counts are the numbers next to filter values that show how many products will remain if the customer picks that value, recalculated in real time after each choice. They keep the customer from clicking into an empty page and show the shape of the assortment at a glance. A filter that offers options leading to zero products is a warning sign: its values are not tied to the current selection, which is fixed with attribute dependencies and proper indexing.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Dependent filters are filters whose options depend on earlier choices, following rules that mirror how the products are actually made, so that invalid combinations never appear. Choosing one material or model limits the sizes, finishes or versions that follow. In a hair-care catalog the chain can run from scalp type to the concern to the active ingredient, with validation rules that keep products contradicting an earlier choice out of the results. These rules live in the catalog data, not in the interface.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A multi-select filter is a filter that lets the customer pick several values of one facet at once, such as two brands or three colors, and shows products matching any of them. It is convenient for people and the fastest way for a catalog to multiply URLs, since every pair or trio of values can produce its own address. Such combinations are almost never what anyone searches for, so their URLs are usually kept out of the search index.

More: Filters and Faceted Navigation: How Online Stores Create Thousands of Duplicate Pages

A filter landing page is a filter combination that matches real search demand, such as sulfate-free shampoos for hair loss, turned into a proper page with a readable address, its own title, H1 and meta description, and a place in the sitemap. It lets a store rank for specific queries without creating a new category by hand. All other combinations stay out of the crawl, and a combination with no products returns 404, as Google's faceted-navigation guidance updated in December 2025 recommends.

More: Filters and Faceted Navigation: How Online Stores Create Thousands of Duplicate Pages

Zero results is the state when a search query or a filter combination returns no products. For the customer it is not neutral: people who get nothing usually conclude that the store does not have the product, and leave. A well-designed store hides filter values that would lead nowhere, never hides a value already chosen, suggests the nearest ranges or a specialist, and records each case as a zero_results analytics event. Repeated zero results point to missing synonyms, attributes or products.

More: The Website Design Process Most Agencies Show You Is Wrong

A zero-results search log is a record of every search on a site that found nothing, kept without personal data and reviewed regularly. It is the most honest list of what customers want and cannot find: missing products, missing synonyms and attributes people care about that the catalog does not describe. We call it the best catalog brief a store will ever get. Together with the filters customers used and abandoned, it is usually the first input of a product discovery project.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A synonym dictionary is a set of word groups that a store's search treats as equal, so customers find products by their own words, by purpose and by partial names, not only by the catalog's terms. It ties together everyday and technical names, or a brand written in Latin and in local script. Its gaps show up directly in the zero-results log, which makes it a living list rather than a one-time setup. The search on peretz.agency uses synonym groups that work across English, Russian and Ukrainian.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Did you mean is a search feature that tolerates typos and, when a query finds nothing, offers the nearest word from the site's own vocabulary instead of an empty page. It turns a misspelled query into results instead of a dead end, and the suggestion should come from words that actually exist in the catalog. On peretz.agency the order is fixed: the other keyboard layout first, then "Did you mean", then results from the other language versions, with a note saying what was done.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Keyboard layout correction is re-reading a search query typed in the wrong keyboard layout as the word the customer meant: "ctj" typed on an English layout becomes the Cyrillic word for SEO. It is essential for stores whose customers switch between Cyrillic and Latin keyboards, where such queries otherwise return nothing. Spelling suggestions cannot repair a query in the wrong layout, so on peretz.agency this check runs before "Did you mean".

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Autocomplete is the list of suggestions a search box shows while the customer is still typing: queries, categories or products. Good suggestions shorten the path to the product and quietly fix spelling before the search runs. It is often weaker than it looks: Baymard Institute (2021) found that 69% of sites did not offer relevant suggestions for slightly misspelled queries, large retailers with budgets and teams among them. Suggestions should draw on the same synonyms and product data as the full search.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Semantic search is search that matches a query to products by meaning rather than by exact words, connecting the customer's purpose, familiar names and synonyms with the product data. It lets a query describing a need find products whose sheets never use those words. AI models help here, but the results still depend on structured attributes and business rules; on an unstructured catalog semantic search only finds wrong products faster. Laravel 13 added native support for vector queries, which makes it practical in custom stores.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Search ranking is the order in which a store shows results, weighing relevance, attribute matches, availability and the store's own priorities; boosting is lifting chosen products or categories within that order. The line between the two is a business decision: a promoted product must not push an exact match out of sight. The rules belong in the specification next to the search technology: whether out-of-stock products drop to the end or disappear, whether a category or a new collection can be lifted for a season, where relevance stops and promotion begins.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Elasticsearch is a dedicated search engine that keeps its own index of the catalog, built from the store's database, and answers search and filter queries without loading the main database. It pays off when the catalog is large, filters combine many attributes or search must handle synonyms and typos at speed. For a modest catalog, well-structured data and flat indexed tables in the main database are often enough. Added to an unstructured catalog, a search engine only gives faster wrong answers.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Product compatibility, or fitment, is data that says which part fits which model and which accessory works with which device, stored in the catalog instead of explained by phone or chat. Once it exists as data, customers can filter by their own model and the team stops answering the same question by hand. In automotive the chain runs from the VIN to the vehicle's identity, its compatible parts and service; a car-brand platform from 2019 already searched spare parts by VIN, connected to the CRM and inventory.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Multi-attribute matching is finding the products that meet several requirements at once, such as size, material, compatibility, price and purpose, rather than filtering on one attribute at a time. This is how buyers of technical goods choose: a part must fit, be in stock and stay within budget simultaneously. Together with query interpretation, facets, dependencies and ranking, it forms the search and filtering logic layer of a store, and it works only on attributes stored as structured data.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Guided selection, or guided selling, is a step-by-step flow that asks the customer a few questions, checks each combination and shows only the products that actually fit or can be made. It suits goods bought on specifications, dimensions or a precise combination of attributes rather than on looks, where a wall of filters overwhelms the buyer. Behind the interface sit the same dependency rules as in dependent filters, so the questions and the catalog cannot drift apart.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A product configurator is a tool that lets a customer assemble a product from options, such as size, finish, trim or package, and see the result and the price of that exact combination. It is not a filter but a rules engine presented as an interface: valid combinations, price effects, exclusive options, inventory, customer-specific prices and often a quote sent to the CRM. "We need a configurator" is a vague requirement that hides most of this logic. For made-to-measure textiles, the price is calculated from the data model instead of being typed in for each variant.

More: The Customer Sees a Car Catalog. The Business Runs a Lifecycle System.

Product comparison is a tool that puts several products side by side on the attributes that matter, and a step in the buying journey between finding a product and deciding on it. It helps only when attributes are stored as consistent structured data; otherwise the table compares blocks of text. For considered purchases such as jewelry, premium textiles or furniture, comparison sits next to provenance, expert guidance and time to reconsider, rather than pushing for an instant decision.

More: How We Design E-Commerce: From Objects to Experience

Product recommendations are suggestions of alternatives and complementary products, and personalization is adapting them to a customer's viewing history, previous orders and preferences. They are worth adding only where a store has enough data: recommendations built on thin data look random and erode trust in the rest of the catalog. "Users need personalized recommendations" is not a requirement until it says what is personalized, from which data and for which business problem. In luxury, curated recommendations show pieces designed to belong together.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A product catalog is the structured set of everything a store sells: products, their variants, attributes, categories and relationships. Its complexity is set by combinations, not by product count: 5,000 simple products are a flat dataset, while 500 products varying by size, color, material and finish can produce tens of thousands of real combinations. That is why 50 products that need complex configuration can be harder to build than 500 simple ones. Fine jewelry adds stone, metal, size, engraving, certification and one-of-a-kind items.

More: What Actually Makes E-Commerce Development Expensive

An SKU, or stock keeping unit, is the store's own unique code for one sellable item, such as a specific size and color of a product, used to track it in stock, orders and reports. It sits in the product data next to the global identifiers GTIN and MPN. Every launch variant needs its SKU before catalog-dependent development starts. SKU count says little about complexity, but SKU quality matters: in one acquisition audit, records full of errors forced a full re-parse of the catalog.

More: What Actually Makes E-Commerce Development Expensive

A GTIN, or Global Trade Item Number, is the standard identifier behind a product's barcode, and an MPN is the manufacturer part number; together they let marketplaces, Google and AI assistants identify an item unambiguously. A store adds them wherever they exist, next to its own SKU. Each variant carries its own GTIN: in our AI shopping agent test, one large US store described 41 variants of a single shoe, each with its own GTIN, size, color, price and availability.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

A product variant is one sellable version of a product, defined by size, color, material or another option. Variants are the multiplier of a catalog: pricing, inventory, search, filters, photography and the admin must agree on which combinations exist, and that logic cannot be finished while the business is still deciding. Google has supported variant markup (a product group with variants) since February 2024, with each variant on its own URL and a canonical to the main product. A color with its own search demand can earn an indexable page; a size rarely can.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Product attributes are the characteristics of a product, such as size, material, color or power, stored as structured fields with allowed values rather than in paragraphs or inside images. Stored that way, they let customers, filters, feeds and machines read the same facts. The same characteristic written three ways by three people breaks filters, and filters cannot be tested against attributes still being invented. Best sellers deserve hand-written descriptions; the long tail can get careful templates filled from real attribute values.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A product data model is the structure of a store's product information: taxonomy, attributes and their allowed values, variants and relationships between products. Each fact is stored once and reused on the product page, in structured data, feeds and internal systems, with standard facts such as price, availability and dimensions kept apart from editorial text. A variant-heavy catalog is a different data model, not a bigger catalog: one product with three variants sold in eight markets already has 24 commercial states.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Catalog data normalization is agreeing one list of attributes and allowed values and cleaning the catalog to it, so that the same characteristic no longer appears under several names or in several formats. In large catalogs it is the bottleneck between raw supplier data, with inconsistent formats and missing specifications, and a publishable product card. Without it, filters show duplicate values and search misses products. Improving the interface while the data stays messy only makes the mess more visible.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

A PIM, or product information management system, is software that owns a company's product information, such as attributes, descriptions, media and documents, and feeds it to the online store, marketplaces and partners. It is a separate type of system from a CRM, which holds customers, and an ERP, which holds stock and finance. A PIM earns its place when several channels and teams work with the same catalog and spreadsheets no longer keep it consistent.

More: Digital Architecture: How to Build a System That Can Grow With Your Business

Product data readiness is the state in which a store's product content is final enough to build on: names, descriptions and prices, launch variants each with an SKU or stock reference, consistent images, decided categories and attributes, and specified bundles. This work sits on the client's side, and it is the most common reason store launches slip. Filters, imports and other catalog-dependent development cannot be finished without it.

More: How Long Does an E-Commerce Build Actually Take, and What the Client Has to Do

Product data quality is how accurate and consistent a store's titles, prices, availability, variants, shipping and return terms are everywhere the product is published: the page, the markup, the feeds and checkout. AI assistants do not need persuasive copy, they need facts they can repeat, which turns this "boring infrastructure" into a strategic asset in agentic commerce. Mismatches are common even among large stores: in our test of 15 US online stores, two showed a checkout price different from the product page.

More: SEO Gets You Found. AEO Gets You Recommended. Your Commerce Infrastructure Gets You Bought.

A catalog admin is the back-office interface where a team creates and edits products, and in large catalogs it is the part that fails first. When it cannot handle the volume, teams retreat to spreadsheets, where data quality dies. A working catalog admin needs bulk editing with validation, staged review before publishing, a change log and a way to reverse a change without restoring a backup.

More: Multi-Country E-Commerce for Large Catalogues: The Architecture Decisions That Set the Ceiling

A product collection is a group of products that belong together by design, material or story, such as a season's line or a set of pieces for one room. A product belongs to a collection, and the collection can carry its own story and page. Collections hold up best when they are built from real material and design relationships in the data rather than from tags assigned by hand, so they follow the range when it changes. On Shopify, collections sit under a fixed /collections/ path.

More: How We Design E-Commerce: From Objects to Experience

A product page, or PDP (product detail page), is the page of one product with its photos, description, attributes, price, availability and buy button, built as a structured, verifiable description for people, search engines and AI. The order of information decides the sale: what it is, whether it suits me, what it is made of, price and delivery, what others say, what goes with it. A sticky buy block keeps the price and button in view, and heavy content such as lab reports or 360-degree views loads only when the customer reaches it.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Product reviews are customers' ratings and comments published on the product page, and described in structured data where they exist. They answer what a description cannot: how the product behaves in real use. Negative reviews are part of that trust: in one pet store case they were published and each unhappy customer was contacted first, and only reviews from unreachable authors and openly abusive ones stayed unpublished.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Product Q&A is a section of the product page where customers ask questions and the store answers them in public. Each answer serves every later buyer with the same doubt and adds specific, useful text to the page. In one pet store case it turned the product page into a support channel, answered by a sales manager who had worked as a veterinarian.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Cross-selling is offering products that go with the one being viewed or bought, such as a matching accessory, a compatible part or the rest of a set. It works when it is built on real relationships (same collection, material, compatibility, an alternative at another price) and turns into noise when everything links to everything. It matters for margins: a store's profit often comes from related products in the same basket and from repeat orders, not from the first promotional order.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Upselling is offering the customer a more expensive, larger or better-equipped version of what they are considering. In mass retail it is a standard lever on order value. In luxury it backfires when it sounds like "spend more"; a curated suggestion that says "these were designed to exist together" fits the category better. The difference lies in the tone and in whether the suggestion comes from real product relationships.

More: Why Luxury E-commerce Breaks Every Rule of Conversion Optimization

A product bundle is a named set of products sold together, with a one-line description, a price and one "Add all" button. Done well, it raises order value without a discount, because it saves the customer the work of putting the set together. Bundles need to be specified before launch as part of product data readiness, with their components, price logic and stock rules, since a bundle is available only while every component is.

More: 10 Best Online Store Designs in 2026 and What Makes Them Work

A considered purchase is one where the buyer evaluates the product and needs to trust the seller before paying: jewelry, premium textiles, furniture, real estate or B2B equipment. It needs provenance, specifications, detailed photography, comparison, expert guidance or a consultation, and time to come back; the catalog, cart and checkout path covers only part of that journey. It is the opposite of an impulse purchase, and the channel follows: impulse items under $50 convert better on TikTok, considered purchases on Instagram.

More: What Actually Makes E-Commerce Development Expensive

Price on request is showing an invitation to ask instead of a number, followed by a clear next step: an enquiry, an appointment or an order by phone. For a $100,000 piece it can fit better than a price next to Add to Cart; at one watch retailer it raised calls and boutique visits by 17% and conversion by 8%, while a discount store needs the opposite. Removing the price without a next step only creates uncertainty. Among nine high jewelry sites, prices were handled in five different ways.

More: Why Luxury E-commerce Breaks Every Rule of Conversion Optimization

Artificial urgency is sales pressure created by countdown timers, exit-intent pop-ups, low-stock badges and loud discounts. It works on $120 sneakers and creates suspicion on a $100,000 piece. Real scarcity, such as a limited edition, a unique stone or a numbered piece, can be meaningful when it is true and explained; invented scarcity undermines a high-value product and the brand behind it.

More: Why Luxury E-commerce Breaks Every Rule of Conversion Optimization

A 360-degree product view is an interactive image that the customer rotates to see a product from every side, often combined with extreme close-ups, video or a 3D model. In luxury and technical goods it gives part of the in-store look that flat photos cannot. Such heavy visuals should load only when the customer reaches that section, so the catalog and the first screen stay fast.

More: Why Luxury E-commerce Breaks Every Rule of Conversion Optimization

Luxury e-commerce is the online selling of high-value goods, such as fine jewelry and watches, by rules that differ from mass retail. Fewer clicks are not always better: the experience may slow the customer down and let them become attached before asking to buy, while urgency, discounts and dense grids work against the product. The website often prepares the customer for the boutique rather than replacing it. Order values change the economics: one premium jeweler's average order is $7,000.

More: Why Luxury E-commerce Breaks Every Rule of Conversion Optimization

Online-to-offline, or O2O, is a sales model in which the website prepares the customer for a visit to a boutique or showroom, where the sale happens, rather than replacing that visit. Design then follows the actual sales model: appointments, store pick-up and consultations matter more than a fast checkout. The weak point is the gap between what happens on the showroom floor and what the platform records, which is where inventory accuracy and customer trust leak away.

More: Why Luxury E-commerce Breaks Every Rule of Conversion Optimization

Clienteling is the long-term personal relationship work of luxury sales associates with their clients: knowing their preferences, sizes and past purchases, and reaching out with pieces chosen for them. In software terms it is a business-specific layer that belongs together with order management, product information and accounting, not a standard CRM function. It sits where online and boutique sales meet, so it needs the same stock and order data as the store.

More: The Software Should Fit the Business

A catalog website is a site that presents products without a cart and checkout, leading visitors to call, write or visit instead of ordering online. For luxury jewelry and watches it is often the right format: a catalog built around the pieces, not a store. Built on a store platform such as OpenCart with the shop features switched off, it keeps the option open, so adding online sales later becomes a small project rather than a rebuild.

More: Why Launching a New Site Is Good for SEO: The Nicole Case

B2B e-commerce is business-to-business online selling, such as a manufacturer selling to dealers or a wholesaler to shops. Compared with retail it adds company accounts with several buyers, customer-specific price lists and catalogs, volume discounts, payment on terms, quotes and repeat orders, often tied to an ERP. These rules are where a standard store platform most often needs custom work.

More: E-Commerce Development

Business rules are a store's own rules for price, delivery, discounts, availability and approvals: a price set by contract, delivery priced by city, a discount that combines with some offers and not with others. They decide which platform fits: test each candidate on your three most inconvenient rules (how each is built, what it costs to maintain, what happens when it breaks) and prototype the hardest one first. The website displays the result of these rules; it should not invent it. In jewelry, where metal prices move daily and each stone is priced on its own, a standard price field stops being enough.

More: E-Commerce Development

Customer-specific pricing is pricing that depends on who is buying: the account, customer group, channel, quantity and contract. The store has to answer "which price and permissions apply to this account, channel, quantity and product right now", and "retail minus a percentage" quietly turns into an account pricing system. Unlike a discount code, which gives the same percentage to anyone who has it, wholesale prices are negotiated per account and often tiered.

More: When E-Commerce Becomes a Business System

Tiered pricing, or volume pricing, is a price that drops as the ordered quantity crosses set thresholds, known as price breaks, sometimes differently by product category or season. It is standard in wholesale and rewards larger orders without a separate negotiation each time. Shopify B2B supports quantity rules and price breaks on all plans; on a custom platform the tiers sit in the pricing engine together with customer-specific prices.

More: When E-Commerce Becomes a Business System

Wholesale e-commerce is online selling in bulk to businesses at negotiated or tiered prices, often on payment terms such as net 30 or net 60. It differs from retail in accounts, prices, order rules and invoicing, not only in order size. Bolting a wholesale section, a minimum order quantity field and a password-protected catalog onto a retail store solves the visible problem and creates three invisible ones: in pricing, in orders and in accounts.

More: When E-Commerce Becomes a Business System

A price list in B2B e-commerce is a set of prices assigned to a customer group, such as retail, wholesale or distributors, often with its own selection of products. On Shopify a few lists fit any plan (up to three B2B catalogs below Plus), while a separate list for each of fifty wholesale clients means Shopify Plus. On a custom platform, price lists live in the pricing rules, as many as the business needs.

More: Shopify Development

A company account in B2B e-commerce is a customer account that represents a business rather than a person: several users share the company's credit line, prices and payment terms, sometimes across several locations. A partner account is a similar entity with its own users and pricing agreement. Without this model, wholesale buyers end up as separate retail customers with prices adjusted by hand. Shopify B2B offers company accounts with locations on all plans.

More: When E-Commerce Becomes a Business System

Net payment terms, such as net 30 or net 60, mean that a business customer pays an invoice within the agreed number of days after delivery instead of paying by card at checkout. They are a normal part of wholesale, so the store has to issue invoices, track what is due and decide which accounts qualify. Order routing applies net terms to wholesale orders only. Shopify B2B supports net terms, and Plus adds deposits and partial payments.

More: When E-Commerce Becomes a Business System

A quote and approval workflow is a B2B buying process in which the customer requests a price, the order passes the buyer's internal approval and becomes a purchase order, followed by recurring orders and invoices. Instead of a cart and instant payment, the store needs quote requests, statuses, approvers and documents. On the purchasing side the same logic covers suppliers, price lists, approvals, receipts and the gap between what was ordered and what arrived.

More: The Software Should Fit the Business

Multi-market pricing is setting prices for several countries as a derived value with a defined order of precedence: base price, market margins, rounding rules and overlapping promotions. Each market has its own taxes, delivery costs, risks and payment commissions, which belong in the price and should be explained before the customer asks. Rounding differs by culture and price bracket, since straight currency conversion produces figures nobody would print. Sales tax, exemptions and shipping rules become part of the architecture, not a step added at checkout.

More: Multi-Country E-Commerce for Large Catalogues: The Architecture Decisions That Set the Ceiling

Checkout is the sequence of steps from the cart to a paid order: contact details, delivery, payment and confirmation. The price it shows must be final: in our AI shopping agent test, 2 of 11 stores showed a different price at checkout, one of them adding a $1.75 add-on nobody had chosen. A good checkout is a separate page without distractions, groups information into clear blocks and saves details for registered customers, while forced registration, long forms and fees revealed on the last screen raise what we call the price of yes.

More: We Sent an AI Agent Shopping in 15 US Online Stores. Here Is Where It Gave Up

Cart abandonment is when customers add products to the cart and leave before paying. Typical causes are shipping costs or fees that appear only on the last screen, forced registration and missing payment methods the audience trusts. It is usually a usability problem first and a marketing problem second: if buyers find products but leave at checkout, more traffic only makes each sale more expensive.

More: Traffic But No Leads? Your Customer Is Doing Math You Never See

A payment gateway is the service that accepts card and wallet payments for an online store and passes them on for processing. On a custom Laravel store the provider is connected directly through its API; on Shopify a third-party gateway carries an extra platform fee, from 2% on Basic down to 0.2% on Plus. Payments must be tested with real transactions before launch. If the method an audience trusts is missing, those buyers are lost however good the rest of the store is.

More: Moving to Shopify Plus? 8 Things to Check Before You Migrate

Multi-currency checkout is charging customers in their local currency, which is not the same as only displaying converted prices. On Shopify, checkout in local currencies works with Shopify Payments or, for enterprise merchants, Adyen; with other providers the charge converts back to the store currency. A currency and region switcher in the header also tells international buyers, before they look for it, that the store ships to them.

More: Moving to Shopify Plus? 8 Things to Check Before You Migrate

An OMS, or order management system, is the system that collects orders from every channel (the website, marketplaces, stores, wholesale) and routes them to fulfilment, invoicing and returns. Its order model must fit how the business actually sells: a system with a single concept of an order cannot represent a wholesale invoice with payment terms or a dropship order tied to a partner. In larger businesses it stands as a separate block next to the ERP, the PIM and accounting.

More: When E-Commerce Becomes a Business System

Order routing is the logic that sends each order down the right path after the customer presses the same "Place order" button: ship from the warehouse for retail, create a purchase order for a dropship supplier, apply net terms and an invoice for wholesale. It can also pick a warehouse by stock and destination. Without it, someone sorts orders by hand, and every new channel adds manual work.

More: When E-Commerce Becomes a Business System

Returns are the process of taking products back, refunding or crediting the customer and putting stock where it belongs, and the rules differ by channel. Retail usually means refund and restock, wholesale often partial credit or a restocking fee, and dropship returns go through the original supplier. Return and shipping policies should also appear in product structured data: in our test of US stores, only 1 of 13 described its shipping terms in machine-readable form.

More: When E-Commerce Becomes a Business System

Click and collect is ordering online and picking the order up in a store, boutique or pick-up point instead of having it delivered. It needs accurate stock per location and a clear signal when the order is ready. On luxury jewelry sites the customer often chooses delivery or boutique pick-up right on the product page, which brings the online buyer into the boutique.

More: We Measured Nine Luxury Brand Websites. Here Is What Their Design Gets Right

A made-to-measure order is an order for a non-standard size, finish or configuration, produced for one customer. It changes material use, price and lead time, so the configuration logic should live in the data model and the price be calculated from the material needed rather than typed in for each variant. Jewelers often create custom orders on the spot in the showroom, which many standard platforms cannot record.

More: Intelligent Product Discovery: Search, Filters and Configurators for Complex Catalogs

Inventory management is tracking stock through its real states: available, reserved, sold, in transit, in preparation, ready for delivery and delivered, because available does not mean ready and sold does not mean delivered. It is usually part of an ERP. Good allocation keeps physical, reserved and channel-available stock as related figures rather than one shared number. For a jeweler with one-of-a-kind pieces, consignment and custom orders, inventory becomes the operational core, and platforms built around "SKU, price, add to cart" handle it poorly.

More: The Customer Sees a Car Catalog. The Business Runs a Lifecycle System.

Inventory synchronization is keeping stock figures the same across the website, the showroom, marketplaces, partners and fulfilment or repair systems. When it fails, the numbers stop matching and overselling follows. It is one of the requirements that push a business beyond a basic CMS, and partners expect real-time stock in a clean feed rather than spreadsheets sent by email.

More: Why Family-Owned Jewelry Stores Outgrow Their E-commerce Platform

Overselling is selling more units than actually exist, and stock reservation is the rule that prevents it by setting units aside for an order. It is rarely a bug of scale: it comes from an undecided reservation model, or from two channels selling the same stock in the same week. The model has to say when stock is reserved, for how long, and what happens when two markets sell the last unit in the same minute.

More: Multi-Country E-Commerce for Large Catalogues: The Architecture Decisions That Set the Ceiling

Multi-warehouse inventory is stock kept and counted in several warehouses, with the store showing availability and delivery options based on where each unit is. One warehouse serving eight markets and eight warehouses serving eight markets are different systems, and the store must decide what to show when a product is available in one country but not in another. Builds with ERP integration and several warehouses take considerably longer than stores with a single stock.

More: Multi-Country E-Commerce for Large Catalogues: The Architecture Decisions That Set the Ceiling

Out-of-stock and discontinued products are items a store cannot sell right now or will not sell again, and each of these states needs one rule for the whole catalog. Temporarily out of stock: the page stays live, the markup says OutOfStock or BackOrder, and links lead to close alternatives. Discontinued with a successor: a permanent redirect; with nothing comparable: 404 or 410, unless the page still helps buyers; an empty category stays out of the index until products return. Site search needs its own rule: out-of-stock items drop to the end or disappear.

More: E-Commerce SEO

An ERP, or enterprise resource planning system, is software that runs the internal side of a business in one database: inventory, purchasing, production, logistics, finance and staff. A CRM looks outward at customers, while an ERP looks inward at resources. Companies usually reach for one when spreadsheets and disconnected tools can no longer keep up. Our custom ERP projects start from $50,000.

More: ERP Development

A CRM, or customer relationship management system, is software that keeps every contact, deal, conversation and task with customers in one place, so that sales and service teams see the full history and the pipeline. Off-the-shelf CRMs cover most standard sales processes; a custom CRM makes sense when the process itself is what sets the business apart. Our custom CRM projects start from $30,000.

More: Custom CRM Development

A POS, or point-of-sale system, is the software and hardware that record sales in a physical store: the till, card payments, receipts and stock movements. In a business that also sells online, it must exchange data with the storefront, ERP, CRM, accounting and shipping, or stock and customer histories split in two. On Shopify, POS, social channels and marketplaces connect to the same catalog and orders.

More: Why Family-Owned Jewelry Stores Outgrow Their E-commerce Platform

A hosted, or SaaS, e-commerce platform is a store platform run by its vendor for a monthly subscription plus transaction fees, with Shopify as the main example. It takes servers, core updates, card compliance and traffic peaks off the merchant's hands, and a themed store can launch in weeks. The trade-offs: a percentage fee stays a percentage as revenue grows from $5 million to $20 million a year, and template workflows flatten what makes a business different.

More: What Actually Makes E-Commerce Development Expensive

Shopify is a hosted e-commerce platform and the faster, cheaper start for standard direct-to-consumer retail in one or two markets. Its US plans (October 2026) are Basic at $29/$39, Grow at $79/$105 and Advanced at $299/$399 a month; they differ in staff accounts, payment fees and how far checkout and B2B can be changed. Products and collections live under fixed /products/ and /collections/ paths. Our Shopify stores start from $10,000 and usually take four to six weeks.

More: Shopify Development

Shopify Plus is Shopify's plan for larger and fast-growing merchants. It starts from $2,300 a month and adds a fully customizable checkout, unlimited B2B catalogs and staff accounts, up to nine expansion stores and higher API limits (Shopify, October 2026). For a standard direct-to-consumer store it is often the simplest choice; business rules unique to the company may need apps or workarounds.

More: Shopify Development

Shopify Markets is Shopify's tool for selling to several countries and regions from one store, each market with its own currency, language, prices and domain. It is available on every plan with no limit on the number of markets; Basic, Grow and Advanced support up to 20 languages, Plus up to 30. Displaying local prices is not the same as charging in local currency, which depends on the payment provider.

More: Shopify Development

Shopify B2B is Shopify's set of wholesale features: company accounts with locations, quantity rules, price breaks, net payment terms and checkout through draft orders. Basic, Grow and Advanced include it with up to three B2B catalogs; Plus adds unlimited catalogs, deposits and partial payments. It covers standard wholesale well, while separate prices for dozens of clients push a business to Plus, and unusual approval or pricing logic toward a custom build.

More: Shopify Development

Shopify Functions are pieces of custom logic for discounts, delivery and payment options that run inside Shopify's own infrastructure. They work under strict limits on instructions, input size and memory, with limited network access and no randomness or clock, so complex rules may not fit. Public apps that use functions work on any plan; custom apps with their own functions require Shopify Plus.

More: Shopify Development

Shopify checkout extensibility is the way Shopify's checkout is customized today: with apps and extensions in set places, not by rewriting the checkout code. Apps on the information, shipping and payment steps are available only on Plus. The old checkout.liquid was switched off on the thank-you and order-status pages on 28 August 2025, so earlier customizations have to move to extensions.

More: Shopify Development

Liquid is Shopify's template language for store themes, and Online Store 2.0 is its theme architecture of sections, blocks, metafields and metaobjects (structured custom data), which lets a team edit pages without a developer. Much of a Shopify developer's routine lives in this layer: theme sections, app integration, fixes and speed. Stores often inherit custom themes and private apps nobody can safely change, and apps added for a campaign can leave scripts in the theme that slow it down even after they are uninstalled.

More: Hire Shopify Developers

Hydrogen is Shopify's framework for headless stores: the storefront is built as a separate application, while Shopify keeps running the catalog and orders and checkout stays within Shopify's checkout tools. It gives full freedom over the storefront at a higher build and maintenance cost. It makes sense only when the storefront and the commerce logic need to change separately, which is why we propose it rarely.

More: Shopify Development

Shopify Payments is Shopify's own payment processing, and Shop Pay is its accelerated checkout that many buyers have already used in other stores. US card rates fall from 2.9% + 30 cents on Basic to 2.25% + 30 cents on Plus, and using it avoids the extra fee Shopify charges for outside gateways. A 1.5% currency conversion fee is not returned when an order is refunded.

More: Shopify Development

App sprawl is a store's growing dependence on many third-party apps or extensions, each solving one problem and adding its own fee, scripts and risks. Fifteen $50-a-month apps do not feel large until there are fifteen, and each can change its pricing, stop being maintained or break after an update. When business rules are spread across apps, they become the permanent workaround that makes a platform expensive and hard to leave.

More: What Actually Makes E-Commerce Development Expensive

Magento, now Adobe Commerce, is an enterprise e-commerce platform with real flexibility and a high cost of staying current. Magento Open Source is free; Adobe Commerce costs roughly $22,000 to $125,000+ a year depending on revenue. Standard support for 2.4.6 ended on 11 August 2026 and for 2.4.7 ends on 31 May 2027, so many stores face an upgrade-or-leave decision. Business rules tend to spread across extensions, XML configuration and plugins; the usual exits are an upgrade, Shopify Plus or a custom Laravel build.

More: Leaving Magento in 2026: Upgrade, Shopify Plus or Custom Laravel?

Hyvä is a front-end theme for Magento that replaces the heavy default storefront with a much lighter one, and it is the most popular way to make a Magento store fast. It became free and open source in November 2025, which removed the licence cost from the decision. It speeds up what customers see without changing the back end, so it does not remove the cost of keeping Magento itself current.

More: Leaving Magento in 2026: Upgrade, Shopify Plus or Custom Laravel?

WooCommerce is the e-commerce plugin for WordPress that turns a WordPress site into an online store. It handles e-commerce well for a few dozen or a few hundred products, and suits businesses whose content already lives in WordPress. With a tight budget it can stretch further: in one case it was chosen over a Laravel build for about 8,000 products with customer accounts and CRM integration.

More: When Does a Business Outgrow WordPress?

OpenCart is an open-source PHP e-commerce platform with no platform fee, suited to mid-size catalogs with standard rules. It is extended with modules and custom code, while hosting and updates stay on the store's side, and modules age when nobody updates them. An OpenCart store does not always need replacing: one hair-care store moved from PHP 5.6 to OpenCart 3.0.5.1 on PHP 8.4, and its filtering is being extended rather than rebuilt.

More: E-Commerce Development

A custom e-commerce platform is a store built around a business's own rules, such as contract prices, B2B terms, several markets and many integrations, usually on a framework like Laravel, with each rule living in one place in the code. There is no platform fee or app stack to rent, but it needs a team to maintain it. One client rebuilt on Laravel saved about $17,000 in the first year compared with Magento. Our online store development starts from $25,000.

More: E-Commerce Development

Headless commerce is an architecture in which the storefront, the part customers see, is built separately from the commerce engine that holds products, prices, orders and payments, and the two exchange data through application programming interfaces (APIs). It allows custom experiences, several storefronts and new channels such as apps or AI interfaces without rebuilding the back end. The cost is more code to build and maintain, so it should solve a real business need.

More: What Happens to E-Commerce When a Website Stops Being a Website?

Composable commerce is an architecture assembled from separate best-of-breed services, such as search, catalog, checkout and content, connected through APIs instead of one platform. Each part can be chosen and replaced on its own. The trade-off is that integration complexity and ownership are spread across many vendors, while a company's own domain layer keeps full control of what makes the business different.

More: The Software Should Fit the Business

Replatforming is moving an online store or a site from one platform to another, for example from Magento to Shopify Plus or to a custom Laravel build, while keeping products, customers, orders, URLs and search rankings. Most of the risk lies outside the new design: data migration, redirects, integrations, checkout and structured data. A written checklist and a redirect map prepared before launch decide whether the traffic survives.

More: Moving to Shopify Plus? 8 Things to Check Before You Migrate

Multichannel e-commerce is selling through several channels, such as retail, wholesale and dropship or reseller partners, from one catalog and one source of truth for stock. Each channel is a different transaction model, so the storefront becomes one interface into a larger operational system. Specifications and media are kept once and presented per channel: rich for retail, specification-first for wholesale, a data feed for partners.

More: When E-Commerce Becomes a Business System

Multi-country e-commerce is selling in several countries through a set of regional stores that share one catalog, each with its own checkout logic, prices, taxes, local delivery and payment providers. Languages do not scale by simple multiplication: three languages are not three times the work of one, because each version must be localized, reviewed and maintained. SEO also needs localized structures for each market, not translated copies of one.

More: Multi-Country E-Commerce for Large Catalogues: The Architecture Decisions That Set the Ceiling

DTC, or direct-to-consumer, is a brand selling to consumers through its own store rather than through retailers or marketplaces. The brand keeps the margin, the customer data and control of the experience, and takes on traffic, fulfilment and service itself. For a standard DTC store in one or two markets, Shopify or Shopify Plus is often the simplest choice.

More: Leaving Magento in 2026: Upgrade, Shopify Plus or Custom Laravel?

Dropshipping is a model in which the seller never touches the product: the order goes to a supplier, who ships it straight to the customer. For the store it requires real-time visibility of the supplier's stock, the partner's own pricing and an order flow that creates a purchase order for the supplier automatically. Returns follow the same route back through the original supplier.

More: When E-Commerce Becomes a Business System

A marketplace is a platform where many sellers sell to buyers, such as Amazon, or a multi-vendor platform built for one business and its partners. AI shopping already leans toward the big ones: only 6.3% of recent AI-assisted purchases were completed on a brand's own site, against 59.7% on Amazon. Building a marketplace means catalog access for dozens of sellers, synchronized orders, centralized inventory and controlled pricing.

More: We Sent an AI Agent Shopping in 15 US Online Stores. Here Is Where It Gave Up

A product feed is a structured file or data stream of a store's catalog, with prices, availability, variants, shipping and returns, sent to Google Merchant Center, ad platforms, AI assistants and partners. Its values must match the product page and its structured data. Feed quality sets the ceiling for Shopping campaigns, and partners want a clean feed with real-time stock instead of spreadsheets by email.

More: Product Page SEO: Descriptions, Reviews and Structured Data That Earn the Click

Google Merchant Center is Google's account where a store uploads its product feed and manages how its products appear across Google. Through free listings, products can show at no cost in Search, the Shopping tab, Images, Lens, YouTube, Maps and Gemini, though without any guarantee. Product markup on the site and a Merchant Center feed together give the widest eligibility. Merchant Center also defines the landing page for Shopping ads, and a healthy feed positions a store for Google's Universal Cart.

More: E-Commerce SEO

The Google Shopping Graph is Google's knowledge base of products, sellers, prices, reviews and stock, which by Google's statement holds more than 50 billion listings, about 2 billion of them updated every hour. AI Mode answers shopping questions from it, and Universal Cart, announced on 19 May 2026, follows a shopper across Search, Gemini, YouTube and Gmail. A store enters it through its feed and structured data, so their accuracy shapes how its products are described.

More: SEO Gets You Found. AEO Gets You Recommended. Your Commerce Infrastructure Gets You Bought.

Agentic commerce is shopping in which an AI agent searches, compares and adds products to the cart, and sometimes pays, on a person's behalf. For a store it means that product data, prices, shipping and returns must be readable by machines, and that bot protection must not block legitimate assistants. In our October 2026 test, an agent reached the cart in 11 of 15 large US online stores. The data work that makes filters reliable for people is the same work that makes a store readable for machines.

More: We Sent an AI Agent Shopping in 15 US Online Stores. Here Is Where It Gave Up

ACP, the Agentic Commerce Protocol, is OpenAI's protocol behind Instant Checkout in ChatGPT, which launched in September 2025 with a 4% fee for merchants. In March 2026 OpenAI shelved Instant Checkout and moved the emphasis to product discovery from merchant feeds, sending shoppers back to the merchant's own site. For stores that keeps the customer relationship with them and makes the quality of their feed the main lever.

More: The Website Is Not Dead. It Has Just Lost Its Job.

UCP, the Universal Commerce Protocol, is Google's protocol that connects AI experiences with merchants' business systems across discovery, shopping and checkout. It is part of Google's move toward assistants that read store data and act on it. For a store, getting ready means the same groundwork as for any shopping agent: an accurate feed, structured product data and consistent prices and policies.

More: The Website Is Not Dead. It Has Just Lost Its Job.

Social commerce is selling directly inside social platforms, where the customer discovers and buys without leaving the app; TikTok Shop is the leading example. It removes checkout friction for impulse purchases and reached $10.3 billion in cumulative US gross merchandise value by the third quarter of 2025. The channel depends on the product: impulse items under $50 convert better on TikTok, premium considered purchases on Instagram, where followers convert 2.7 times as often.

More: TikTok vs Instagram for Small Business in 2026: What the Data (and My Clients) Actually Show

Digital retailing in automotive is buying a vehicle online from start to finish: configuring it, estimating payments, valuing a trade-in, applying for financing, uploading documents, signing electronically and scheduling delivery, without visiting the dealership. It depends on connected dealer systems, pricing layers and inventory rather than on the website alone. The VIN becomes the key that links the customer, configuration, warranty, service history and parts.

More: The Customer Sees a Car Catalog. The Business Runs a Lifecycle System.

AOV, or average order value, is a store's revenue divided by the number of orders over the same period. With revenue, conversion rate and cart abandonment it tells most of a conventional store's story, and in a due-diligence review it sits next to gross versus net sales and repeat purchase rate. It sets how much a store can spend to win a buyer: a premium jeweler with a $7,000 AOV and about 30% margin earns about $2,100 gross per sale, against $150 to $250 to acquire a buyer.

More: Rising CPC Isn't the Problem. Here's What Is

Repeat purchase rate is the share of customers who buy from a store again within a given period. In mass e-commerce it belongs with basket margin and payback time among the numbers that matter most, because profit usually comes from repeat orders, not from the first promotional one. A store that ignores it can grow revenue on paid traffic while losing money on each new customer.

More: Rising CPC Isn't the Problem. Here's What Is

Subscription commerce is a model in which customers receive products on a schedule, with autoship reorders charged automatically, so the business is built on the reorder rather than on the first sale. The first purchase is only the way in; the economics come from customers who stay. At one pet retailer, autoship subscribers made up 83.3% of net sales.

More: Rising CPC Isn't the Problem. Here's What Is