Identity is more than a marketing name
Lasting Index starts with the raw brand and model strings published by the source. Normalisation can make Unicode, spacing and obvious dash differences searchable, but meaningful suffixes and punctuation remain. Those characters can encode capacity, finish, generation, voltage, retailer channel, package contents or market—even when the base name is shared.
A useful exact identity combines category, jurisdiction, brand, complete model code and source record. Dates and other identifiers can strengthen the match, but no single similarity should silently transfer assertions. If an appliance label and catalogue record disagree, the uncertainty should be resolved before comparing specifications or safety evidence.
A registration family is not a universal equivalence class
Australian GEMS rules allow a registration to cover a family of models when the relevant determination’s conditions are met. That proves a defined regulatory relationship inside that registration. It does not prove that an overseas product, later generation or similarly named retailer variant belongs to the same family or shares every non-energy characteristic.
French durability data is also published per model under a specific statutory schema and producer record. A French score cannot be copied to an Australian model because the brand and model stem look familiar. Different legal regimes, calculation evidence, components and market configurations must remain visible.
Why Lasting Index defaults to DO_NOT_MERGE
A merge is high impact: once two identities become one, every attached assertion can appear to describe both. An incorrect merge can move energy values, recalls, dates or statutory scores across jurisdictions. The safe default is therefore to keep candidate relationships separate until an authorised reviewer records evidence for a specific resolution.
The current catalogue retains 722 multi-model relationship candidates as DO_NOT_MERGE. That label does not claim the products are definitely different; it records that equivalence has not been approved. Search and comparison can expose the candidates without collapsing their provenance or presenting one regional record as another.
A disciplined matching workflow
Read the rating plate or official document and transcribe the full model code. Select the market where the product was supplied. Search within the correct appliance category, then compare the returned brand, code, registration or source record and dates. If several records remain plausible, keep them separate and inspect the official source links.
For a recall, also check serial, batch and sale-period scope. For energy data, confirm capacity or volume and registration family. For French durability, confirm the calculation date and producer. A mismatch in any material field should stop automatic transfer and become a review question rather than a guessed correction.
- Preserve the raw model string.
- Lock the evidence to its jurisdiction.
- Keep every assertion attached to its source record.
- Record candidate relationships without merging them.
- Require evidence and an authorised review for resolution.
What this means for search and comparison
Search may return near-looking codes so a person can find the right record, but the model page and comparison tool operate on the selected exact identity. Missing fields are not completed from another result. Comparisons stay inside one market and category, and each value continues to show its own source record.
This approach is less convenient than assuming equivalence, but it makes the uncertainty auditable. It also lets new evidence resolve one relationship without rewriting unrelated records. The goal is not to maximise apparent catalogue completeness; it is to make every published claim traceable to the appliance identity the source actually described.