Lightweight Mode
A streamlined MXP deployment that delivers core search, merchandising, and recommendations through a simplified interface
MXP runs in one of two application modes. General mode exposes the complete platform: multiple tenants, audience segmentation, per-query and per-category facet management, and advanced rule configuration. Lightweight mode exposes a reduced surface of that same platform — the same services, the same APIs, the same catalog — with fewer screens and simpler forms.
Lightweight mode is not a separate product. It is the same MXP deployment with the enterprise configuration layer switched off.
What it solves
A full merchandising platform gives large teams the control they need: several tenants to keep brands apart, audience segments to tailor results per cohort, and rule metadata to keep hundreds of rules organized. For a single-brand deployment with one catalog and a small team, that control surface is overhead — more screens to learn, more fields to fill in, and more decisions to make before anything ships.
Lightweight mode removes that overhead. Merchandisers see the screens they actually use, and the forms ask only for the fields that apply. Configuration is organized around Serving Configurations instead of platform-wide configuration records, so a single object describes how results are served and everything else points at it.
When to use it
- Single-catalog deployments — one brand, one catalog, one team, where tenant separation adds no value
- Embedded consoles — MXP administration surfaced inside another application, where a compact interface matters
- Deployments that do not need advanced rule configuration — teams that boost, bury, pin, and filter, but do not need query rewriting, rule priorities, or audience targeting
- Review and audit roles — users who inspect configuration and metrics rather than author changes
Key concepts
Application mode — the deployment tells the interface which mode it is running in when the interface loads. The mode is fixed by the deployment: it cannot be changed from the UI, and a user cannot switch between modes.
Lightweight Mode indicator — in Lightweight mode the header shows a Lightweight Mode label in the position where the full platform shows the tenant selector. This is the quickest way to confirm which mode you are in.
Default tenant — every request is served against a single default tenant. There is no tenant selector and no tenant switching.
Serving Configuration — the central configuration object in Lightweight mode. Each serving configuration defines how search, browse, or recommendation results are served, and other features reference it: discovery rules, linguistic overrides, and evaluation environments are all scoped to a serving configuration. It takes the place of Global Configuration, which is not available in Lightweight mode.
Facet and Tile Controls — a Lightweight-only screen for adjusting how facet and tile values are returned, replaced, grouped, and ordered. It takes the place of Search and Browse Facet Management.
How it works
Lightweight mode uses the same Merch Module UI as the full platform. When the interface loads, it reads the deployment's mode and adapts in four ways:
| What adapts | In Lightweight mode |
|---|---|
| Navigation | Lightweight-only pages (Serving Configuration, Facet and Tile Controls) appear. Full-platform pages (Global Configuration, Target Groups, Search and Browse Facet Management, Visual Navigation Suggestions) are hidden. |
| Lists | Columns that have no meaning without the enterprise layer are dropped, such as rule tags and priority. Filters for Serving Config ID are added where relevant. |
| Forms | Fields tied to features that are not present are omitted, such as user groups and target groups. |
| Tenant context | The default tenant is applied automatically to every request. |
Because the underlying services are identical, anything configured in Lightweight mode behaves at query time exactly as it would in the full platform. The difference is in what the interface lets you configure, not in how results are served.
What is available in Lightweight mode
This table covers every feature in the Solutions section, ordered from unchanged to unavailable.
| Feature | In Lightweight mode | What that means |
|---|---|---|
| Metrics | Full | All dashboards, date filtering, grouping, count modes, chart and list views, and CSV export. Nothing is removed. |
| Recommendation Models | Full | Listing, creating, editing, and deleting models, plus tune, pause, and resume operations. |
| Linguistic Override | Extended | Adds a Serving Configuration selector on the override form, and list filters by name and Serving Config ID. |
| Autocomplete | Adjusted | Term lists, CSV import and export, and query evaluation are unchanged. Adds Matching Order. Show Top Categories is not available, so suggestions carry no category information. |
| Discovery Rules | Reduced | Boost, bury, pin, filter, redirect, and the diversity and price ranking overrides all work as normal. No Rewrite Query action, and no priority, tags, or description fields. Apply Filter and Boost Attribute take a list of values with an "is any of" operator. The tags and priority columns are hidden, the list filters by Serving Config ID, and rules are scoped to a serving configuration. |
| Recommendation Rules | Reduced | Model ID is the only trigger, and the trigger section is disabled when editing a rule. Boost Attribute is the only action. No Page Type, Products, or User Group triggers, and no Pin Product, Apply Filter, Override Diversity, or Override Price Ranking. Diversification and price reranking move to the serving configuration. |
| Attribute management, part of Configurations | Reduced | A table with inline toggles for Indexable, Facetable, Searchable, Retrievable, Exact Match, and Filterable, plus Modify Controls for bulk editing. No attribute creation or deletion, and no facet ranges, manual facet order, value renaming, or single and multi select option. |
| Evaluation | Reduced | Query evaluation, environment comparison, facets, pagination, filters, and result explanation are unchanged. Adds recommendation testing with a visitor ID field and Add Product. Environments come from serving configurations, the settings button opens the serving configuration, the applied-rules and global-configuration panels are hidden, and link copying is not available. |
| User Management | Reduced | The Users list, search, history log, and the create, edit, and delete actions are unchanged. No Target Groups screen, and roles for full-platform-only features are not offered in the role dropdown. |
| Approval Requests | Reduced | Queues, filters, side-by-side review, approve and reject, cancellation, and history are unchanged. Serving Configuration and Facet and Tile Controls approvals are Lightweight-only; Global Configuration, Target Group, facet management, and Visual Navigation Suggestions approvals do not arise. |
| Serving Configuration | Lightweight only | Defines how search, browse, and recommendation results are served. |
| Facet and Tile Controls | Lightweight only | Ignore, replace, remove, merge, rerank, and force-return facet and tile values. |
| Global Configuration, part of Configurations | Not available | Replaced by Serving Configuration. |
| Target Groups, part of Configurations | Not available | No audience segmentation. |
| Search Facet Management | Not available | Facet behavior is managed in Facet and Tile Controls instead. |
| Browse Facet Management | Not available | Facet behavior is managed in Facet and Tile Controls instead. |
| Visual Navigation Suggestions | Not available | Image-driven navigation suggestions require the full platform. |
| Dynamic Categories | Not available | Virtual category pages built from saved queries require the full platform. |
| Recommendation Containers | Not available | Placement management requires the full platform. In Lightweight mode a recommendation serving configuration points at a model. |
| Attribute Enrichment | Not available | AI-generated attribute values, review and publish, and enrichment runs require the full platform. |
| Lexical Mining | Not available | Requires the full platform. |
| Experiment Metrics | Not available | Experiment-level reporting requires the full platform. Standard Metrics dashboards are fully available. |
| Multi-Tenancy | Not available | A single default tenant, with no tenant selector. |
Every feature page carries a note pointing back to this table, so whichever page you land on tells you what applies to your deployment.
What the full platform adds
Lightweight mode covers day-to-day merchandising. Teams that need more control move to General mode, which adds:
- Multi-tenancy — run isolated brands, business units, regions, or environments in one deployment, and switch between them from the header
- Global Configuration — platform-wide search and facet defaults, a default facet list, persistent boost, bury, and filter conditions, and version history with rollback to any prior state
- Target Groups — named audience segments with priorities and context paths, each able to receive its own configuration, for VIP shoppers, internal users, or A/B cohorts
- Search and Browse Facet Management — per-query and per-category facet control, configured independently for search and browse surfaces
- Visual Navigation Suggestions — image-driven suggestions alongside text autocomplete
- Full attribute management — create and delete attributes, plus advanced facet configuration: facet ranges, manual facet ordering, value renaming, and single or multi select behavior
- Advanced rule configuration — query rewriting, rule priorities, tags, and descriptions on discovery rules, and page type, product, and user group triggers with pin, filter, diversity, and price ranking actions on recommendation rules
- Richer evaluation — shareable evaluation links, per-environment parameter management, and the applied-rules and global-configuration panels
- Attribute Enrichment — AI-generated values for missing or incomplete product attributes, with a review and publish workflow and a record of every enrichment run
- Dynamic Categories — virtual category pages defined by saved search queries, which stay current as the catalog changes instead of being manually curated
- Recommendation Containers — map one or more models to a specific placement, such as a product detail or listing page
- Lexical Mining — vocabulary and term extraction from catalog and query data
- Experiment Metrics — reporting scoped to individual experiments, on top of the standard Metrics dashboards
The application mode is set when the environment is deployed, not from the interface. To find out which mode your environment runs, or to discuss moving to the full platform, contact your MXP administrator or Grid Dynamics support.
Related pages
- Serving Configuration — the central configuration object in Lightweight mode
- Facet and Tile Controls — Lightweight-only facet and tile behavior management
- Solutions — the full catalog of MXP capabilities
- Multi-Tenancy — the tenant isolation model that Lightweight mode replaces with a single default tenant
- Navigating the Platform — a map of the Merch Module UI
Solutions
All MXP capabilities — product discovery, recommendations, attribute enrichment, metrics, evaluation, configurations, approval requests, multi-tenancy, and user management
Serving Configuration
Named configurations that define how search, browse, and recommendation results are served in Lightweight mode