AI · Product data
Structured product data for lighting: the data model behind AI matching
AI is only as good as the catalogue it searches. In lighting, a catalogue is not a list of products; it is a set of rules for building them.
Short answer
AI matching for lighting needs product data in five layers: classified attributes (for example in the ETIM classification: mounting, light source, flux, CCT, CRI, voltage, dimming, IP and dimensions), photometric files (IES LM-63) for delivered output and distribution, compliance data (certification marks, location ratings, DLC listing), ordering logic (which catalogue options combine into a valid product) and commercial data (price, lead time, country of origin). Without the ordering logic and photometrics, a matcher can only compare descriptions.
Key takeaways
- Classified attributes make products comparable across manufacturers.
- IES files carry the performance that catalogue headlines hide.
- Compliance data is pass/fail and must be current.
- Ordering logic turns a family into valid catalogue numbers.
- Commercial data (lead time, origin) now decides as often as specs do.
Five layers of lighting product data
| Layer | Contents | Why AI needs it |
|---|---|---|
| 1. Classified attributes | Mounting, light source, flux, CCT, CRI, voltage, dimming, IP, dimensions | Compare products across manufacturers |
| 2. Photometrics | IES LM-63 files per configuration | Delivered output and distribution |
| 3. Compliance | Certification marks, location ratings, energy listings (DLC) | Pass/fail constraints |
| 4. Ordering logic | Valid option combinations, lengths, increments | Propose orderable products |
| 5. Commercial | Price, lead time, country of origin, stock | Decide between equals |
1. Classified attributes
Standards such as ETIM give each product class a fixed list of features and allowed values: numeric (power, voltage, length), lists (colour, material, mounting) and yes/no. For luminaires that covers mounting method, light source, luminous flux, colour temperature, colour rendering, control gear and dimmability, protection ratings and dimensions. Classified data is what lets a system compare a recessed linear from one manufacturer with another’s.
2. Photometrics
Catalogue lumens and “wide/medium/narrow” are not comparable. IES files describe how much light leaves the fixture and where it goes. Matching on photometrics is what makes an alternate defensible in a substitution.
3. Compliance
Certification for Canada, location rating and energy listing are pass/fail and change over time: listings are added and removed. See certification marks in Canada and DLC listing.
4. Ordering logic
A linear family may offer dozens of lengths, outputs, optics, CCTs, voltages and finishes. Not every combination exists; some depend on others. Encoding the ordering rules lets the system output a valid catalogue number and flag open options, rather than a string that looks right.
5. Commercial data
Lead time, country of origin and price now decide as often as performance. Tariffs and component shortages make them volatile. See AI for supply-chain risk.
From catalogue to knowledge graph
Linked together, these layers form a product knowledge graph: families, configurations, photometric files, listings and commercial terms, connected so a query like “347V, DALI, 4000K, CRI 90, recessed trimless, under 5 weeks” returns valid, comparable answers. That graph is what AI product matching searches.
Why it is a moat
Models are increasingly available to everyone. Clean, current, normalized product data and the record of which matches people approved are not. That is where an AI procurement business builds lasting advantage. See our AI roadmap.
Frequently asked questions
What is ETIM?
ETIM is an international, vendor-neutral classification for technical products. Each product class (for luminaires and other items) has a fixed set of features and values, so manufacturers and distributors can exchange comparable product data.
What is an IES file?
A photometric data file in the ANSI/IES LM-63 format describing how a luminaire distributes light. Design software and matching systems use it to compare delivered performance.
Why is ordering logic important?
Lighting catalogue numbers are built from options (length, output, CCT, optic, voltage, dimming, finish). Not every combination exists. Ordering logic lets a system propose a valid, orderable product rather than a plausible string.
Who owns the product data?
Manufacturers publish it; distributors and platforms normalize it. Keeping it current (new options, discontinued items, listing changes) is ongoing work.
References
Keep reading
- AI · Product matching AI product matching for lighting: how an equal is found and ranked How AI product matching works for commercial lighting: why it is an entity-resolution problem, how candidates are generated and filtered by hard constraints (voltage, dimming, mounting, certification), how equals are scored on photometrics and colour, and why every match needs evidence and a human decision. →
- Compliance · Incentives DLC listing for commercial LED fixtures: when it matters and how to check What DLC listing means for commercial LED fixtures, why most utility rebate programs require it, the difference between DLC Standard and Premium, and how to check a product on the Qualified Products List before ordering. →
- Compliance · Certification CSA, cUL, ETL: which certification marks lighting needs in Canada Electrical products installed in Canada must be certified for Canada by an accredited certification body. What the CSA, cULus and cETLus marks mean on lighting, why a plain UL mark is not enough, and how to check a fixture before it reaches site. →
- AI · Roadmap AI in lighting procurement: a technology roadmap from spec to site A technology roadmap for AI in commercial lighting procurement: reading specifications, matching every line to verified products, assembling quotes and submittals, and the next stages, learning from review, supply-chain intelligence and agentic workflows. Where AI helps today, where people stay in the loop, and what to measure. →