Trusted catalog intelligence
Match comparable products with confidence visible.
Automate candidate discovery, explain the matched attributes and route uncertain relationships to human review before they influence pricing.
Automation proposes. Reviewed evidence decides.97%confidence
Brand exactCategory exactDimensions exactName 94% similar
Clean data can still create a bad decision
A wrong product relationship contaminates every metric downstream.
Matching needs confidence, explanation and review—not a black-box “matched” label.
Similar is not equivalent
Small differences in size, pack, variant or specification can invalidate a price comparison.
Manual matching does not scale
Large catalogs overwhelm teams when every candidate starts from a blank screen.
Uncertainty gets hidden
A binary match without confidence makes risky relationships look final.
Candidate review, not blind linking
Inspect what agrees, what differs and what still needs judgment.
Switch between candidates to see how brand, category, dimensions, fuel type and product name contribute to the relationship. Low confidence never has to become a pricing comparison.
Product matching review_11b_product_matching.html.twig · _product-matching.scss · product-matching.js
Product matching
Review the evidence, not a black-box score.
Candidates, confidence and the attributes behind the relationship remain visible before a comparison reaches pricing.
3 candidates reviewedSelect a candidate to inspect its evidence
Manual reviewMatch confidence97%Confirmed candidate
- BrandExact
- CategoryExact
- DimensionsExact
- Fuel typeExact
- Product name94% similar
Match confidence84%Review recommended
- BrandExact
- CategoryExact
- DimensionsPartial
- Fuel typeExact
- Product name78% similar
Match confidence61%Do not auto-link
- BrandDifferent
- CategoryExact
- DimensionsPartial
- Fuel typeExact
- Product name52% similar
Route uncertainty to the right place
Automation handles volume. Review protects the comparison.
Confidence thresholds can confirm strong candidates, hold ambiguous relationships and prevent weak matches from entering downstream monitoring or pricing.
- Direct competitor links remain supported
- Important categories can require review
- Rejected candidates remain part of the audit context
matching / review queueHefesto Altar grillAll required attributes agree
Auto-confirmHefesto Compact grillDimensions require review
ReviewGarden Pro grillBrand and name conflict
Do not linkRelationship as durable product knowledge
A confirmed match should not lose its evidence downstream.
The reviewed relationship, confidence and relevant attributes travel into the offer table, market position and recommendation. Teams can return to the original basis of the comparison.
- Source identifiers remain traceable
- Match status is separate from collection health
- Manual and automatic relationships are distinguishable
Product relationshipSKU 40218 ↔ GS-40218
ConfirmedCompetitor monitoringOffer becomes comparable
ReadyMarket positionRank uses reviewed relationships
6 / 13Pricing recommendationExplanation retains match context
AuditableDifferent catalogues, different evidence
Match the product definition your category actually needs.
Identifiers can be decisive in one catalogue and incomplete in another. Reevals keeps the comparison logic and review threshold explicit.
Candidate matching
Rank likely competitor relationships using normalized identifiers and comparable attributes.
AssistedConfidence and explanation
See which attributes support the proposed relationship and where uncertainty remains.
97%Human review queues
Prioritize low-confidence, high-impact and newly discovered candidates for review.
ControlledBulk catalog operations
Review related candidates efficiently while preserving per-product evidence.
ScalableConfidence is evidence, not certainty
Assisted matching, never guaranteed perfection.
Confidence thresholds, manual review and relationship types make uncertainty operationally visible.
- 01No forced low-confidence match
- 02No hidden attribute conflict
- 03No downstream use before the configured state
Questions before implementation
The practical questions teams ask first.
The useful answer depends on your catalogue, sources, constraints and ownership. These are the boundaries we establish before automation.
Reevals proposes and scores candidates. Teams define confidence thresholds and can require human review before a relationship is used.
Yes. Relationship types can distinguish exact, equivalent and rejected candidates so pricing strategies can treat them differently.
They are routed to a review queue with the compared attributes and confidence explanation visible.
The reviewed relationship is reused by monitoring, recommendations and analytics so teams do not rebuild the comparison in each module.
Continue the decision loop
Connect the next capability when the workflow needs it.
Product matching
Bring one real decision—not a perfect data set.
We will map the evidence, constraints, ownership and technical boundary required for a useful first workflow.