Illustrative local catalog · no store connected
Turn a buyer question into explicit criteria, then inspect which product fields support the shortlist and which still need evidence.
Four invented garment records and original sketches. This walkthrough filters this finite sample locally; it does not call a search or AI service.
ZIP-01
CREW-02
JKT-03
JKT-04
Choose a preset, then adjust color or budget. Known fields determine the result; missing rain evidence never becomes a match.
Illustrative local catalog · no store connected
Four invented garment records and original sketches. This walkthrough filters this finite sample locally; it does not call a search or AI service.
Known sample matches: 2. Needs evidence: 0.
ZIP-01 · Layer
Sample price: €68.00
Source: invented local sample
Why it is shown: The declared category, color and sample price meet the active constraints.
Material: Cotton blend
Rain suitability: unknown
Availability: not specified
CREW-02 · Layer
Sample price: €52.00
Source: invented local sample
Why it is shown: The declared category, color and sample price meet the active constraints.
Material: Not specified
Rain suitability: unknown
Availability: not specified
These are synthetic examples. No store is queried, no preference is learned and no product is added to a cart.
Selecting a question changes the criteria; it does not fill gaps in a product record. Read known attributes and unresolved requirements separately.
The green-layer question compares the sample category, color and price. A missing material field remains visible even when material is not part of that question.
Neither sample jacket has a rain-suitability specification. The rain question therefore shows them as Needs evidence, without calling either a rain jacket.
Agree what the person needs and which parts are ambiguous.
State category, color, budget and any required specification.
Compare the supplied fields and retain missing requirements.
A person reviews range fit and decides what to investigate next.
This page uses two declared presets and local filtering. It demonstrates no personalization, click learning, visual retrieval or commercial ranking service.
Agree a product and variant key, attribute meanings, locale and currency, availability semantics, image rights and source freshness before a catalog trial.
| Identity and variant | A stable key links the result to the right record. A missing size prevents a size-specific answer. |
|---|---|
| Attributes and units | Define material and weather specifications. An absent value stays unknown; it is not a false or a confirmed match. |
| Locale and currency | Supply the language and currency of the intended market. Price comparisons need consistent units and context. |
| Availability and rights | Define what availability means and agree image rights. A sketch or record here does not establish stock or permission for a real asset. |
| Source and freshness | Retain the source reference and date. A review owner decides whether the record is current enough for the question. |
Compare the known shortlist, inspect range fit and test a proposed assortment with its owner. A filtered sample is not a buying decision.
Continue to Hero ProductsA real trial needs authorized data and agreed evaluation criteria. This illustration establishes no search latency, conversion uplift or sales result.
Only the four explicitly invented garment records shown in this walkthrough. The two presets and color/budget choices filter that finite local sample. No store, AI service or customer account is connected.
They remain unknown. When the question requires an absent field, the record is shown as Needs evidence outside the known shortlist. The page does not infer or repair a product specification.
An authorized sample, stable product and variant identity, agreed fields and market, source dates, asset rights and a review owner. Actual source availability and access are qualified for a workspace before setup.
Agree the sample, required evidence and review criteria before planning a workspace trial.
Discuss a catalog pilotCompare two original sketches and their fictional product fields. These pairs are authored examples, not image-search results.
Four existing fictional records · no image upload or retrieval
Original generic drawing; not photograph, crop or specification evidence.
GarmentSketch/ZIP-01
discoverySample/ZIP-01/sourceDate
Original generic drawing; not photograph, crop or specification evidence.
GarmentSketch/CREW-02
discoverySample/CREW-02/sourceDate
The color and category agree; the product keys and drawn closures differ. Material is missing for CREW-02. Neither record supplies a size.
A drawing does not establish fabric, size, availability or product identity. Different product keys remain different records.
Review product informationA preferred position does not resolve a missing field or a conflicting rule. Review three fictional recommendation cases; no shopper receives these results.
Fixed fictional example — no recommendations served
Switch between fixed examples of inclusion, ordering and missing evidence. These controls change this local explanation only.
Keep the placement, category scope and rule version visible. A rule for one placement is not automatically a rule for the whole store.
Compare each product key with its supplied category and exclusion flag. A missing flag remains unknown, rather than becoming false.
In this example, an ordering preference applies only to candidates whose required fields are known and eligible. This is an illustrative NOGL policy, not a claim about every recommendation system.
P-A meets the category rule but has an explicit exclusion. No candidate is eligible in this example.
Check rule scope, exclusions and missing values before proposing a change. Nothing here approves, saves or publishes a campaign.
Use product information and merchandising review to resolve the open question. The example demonstrates no shopper profiling, learned relevance or commercial uplift.