Product data for tenders: what AI matching needs from your item master

Disclosure: this guide is published by Scalera. Information about Scalera comes from our website; information about standards and regulation comes from the publications of the respective organizations. Please send corrections to info@scalera.ai.

Why does product data decide how well tender matching works?

Because AI can only suggest products whose properties are in your data. A tender line item might ask for a specific fire rating, a thickness and a standard. If one of those attributes is missing from the item master, the product name is the only clue left.

Scalera matches line items based on specification, application context and standard, not just keywords (Scalera, product matching). That logic works better the more complete your attributes are. How the matching itself works is explained in Product matching at scale.

Which product data does AI matching need?

Beyond item number and name, mainly technical attributes with units, standards and classes, units of measure and the logic behind your variants. The table shows which field does what in matching.

Examples for illustration. Basis: matching by specification, application and standard (Scalera, product matching); unit conversion and variants (Scalera, FAQ).

What matters is that attributes exist as separate fields rather than hiding in the long description. A value like “100 mm” in running text is readable for people, but a field called “thickness” with a number and unit is far better for clean matching.

What do ETIM, ECLASS and BMEcat do?

They give product data a shared structure: ETIM and ECLASS define which features describe a product, BMEcat defines how a catalog is transferred. For matching, the structure is what counts, not the format itself.

ETIM: classes, features, values, units

ETIM is an international classification standard for technical products. Each product class groups comparable products from different manufacturers and defines the features that describe them (ETIM International, model information).

ETIM distinguishes four feature types: alphanumeric with a fixed value list, logical (yes or no), numeric, and range with two limits. Numeric and range features always need a unit, and every element carries a language-independent identifier. That is exactly the structure matching against tender requirements needs.

ECLASS: classification across industries

ECLASS is another data standard for classifying products and services. ECLASS e.V. publishes a new release every year (ECLASS). Which standard to use is best agreed with your distributors and customers.

BMEcat: exchanging catalogs

BMEcat is an XML-based standardized exchange format for B2B catalog data; the current version is 2005.2 (BME, BMEcat). The specification includes a dedicated module for classification systems, catalog group systems and feature systems (BME, BMEcat downloads). That way a BMEcat catalog can carry ETIM or ECLASS features.

Industry standards in building materials distribution

In German building materials distribution there is also the DPB industry standard. The association behind it was founded in 2019 by purchasing groups and associations of building materials dealers and aims to simplify master data processes between manufacturers and distributors (DPB Datenmanagement).

Don't confuse these standards with GAEB: GAEB DA XML governs the exchange of construction information for tendering, awarding and invoicing (GAEB, data exchange). It describes the request, not your catalog. How GAEB, ÖNORM and NPK affect quoting speed is covered in GAEB, ÖNORM, NPK.

How should you handle units of measure?

Store, for each item, how its sales unit converts into the units used in bills of quantities. A tender asks for square metres or linear metres, while you often sell in packs, rolls or pieces.

If a conversion factor is in your product data, Scalera converts automatically; otherwise it notifies you and you can add the converted quantity manually (Scalera, FAQ). Without the factor, every affected line item means manual work.

How do you model variants and configurable products?

Describe variants through attributes instead of long item lists with cryptic codes. Then the AI can derive the right version from what the tender specifies.

Scalera works best with standard products with fixed article numbers and configurable products with up to 30 definable attributes. The AI extracts technical specifications from tender documents to match the right product variant (Scalera, FAQ). For products with many degrees of freedom, it pays to clean up the most important selection attributes first.

What role do PIM and ERP play?

If you work with both systems, decide which one leads for technical attributes and which for item numbers, units and prices. For matching, what counts is that both records meet at the same item via the same item number.

Scalera can be integrated with ERP, CRM and PIM systems, and price calculations remain in your ERP or CRM system (Scalera, FAQ). The interfaces available are described in ERP integration for tenders and quotes.

How do you start if your data isn't perfect yet?

With a product line that gets many requests and is well maintained. Many Scalera customers start with their highest-volume or most standardized product lines and expand later (Scalera, FAQ).

After that, the system shows you where data is missing: Scalera uses confidence scores and flags low-confidence matches for review (Scalera, FAQ). If uncertain matches pile up in one product group, that usually points to missing attributes. Scalera also learns with every completed offer which products were offered for which tender line items.

“Scalera is our new employee in processing offers — thinks quickly, never complains, doesn't take breaks and knows every position by heart!”
Marc Caminada, Lazzarini

Why does clean data pay off beyond matching?

Because structured product data will also be required by regulation in the EU. The new Construction Products Regulation (EU) 2024/3110 entered into force on 7 January 2025; its main part has applied since 8 January 2026 (European Commission, CPR-2024).

It introduces a digital product passport that provides machine-readable information about a construction product throughout its life cycle, including the declaration of performance and conformity (European Commission, CPR-2024). The Commission plans the specific requirements for construction products for Q2 2027 (European Commission, Digital Product Passport). If you maintain structured attributes today, the same work serves that purpose later.

Checklist: is your item master ready for AI matching?

Check these points on one product line with high request volume:

1. Attributes as fields: are dimensions, classes and standards in separate fields rather than only in the long text?
2. Units: does every numeric value have a unit?
3. Conversion: are there factors from the sales unit to m², linear metres or pieces?
4. Variants: can versions be told apart by attributes rather than codes?
5. Application: is it clear what each product is suitable for?
6. Classification: are products assigned to a class in ETIM, ECLASS or your own logic?
7. Ownership: is it clear who adds missing attributes when the AI is unsure?

Frequently asked questions about product data for tenders

Do I have to clean up all my product data before starting with AI matching?

No, not completely. Many companies start with their highest-volume or most standardized product lines and expand later. Scalera shows confidence scores and flags low-confidence matches for review, so you see right away where data is missing.

Do I need ETIM or BMEcat to use AI matching?

No, neither standard is a prerequisite. What matters is structured attributes with units. If your products are already classified in ETIM or ECLASS, though, you already have that structure.

What is the difference between ETIM and BMEcat?

ETIM defines which features describe a product class. BMEcat is an XML format for exchanging catalogs between companies. The two complement each other: BMEcat can carry features from classification systems.

Does the AI recognize line items specified with a competitor's product?

Yes. Scalera recognizes both your own and competitor products in tenders and suggests the best matching alternatives from your catalog. How to prove equivalence is covered in Winning with alternates.

Can I keep separate catalogs for countries or business units?

Yes. In Scalera, a location is a segmented team within your account that can have its own product catalogs and workflows. That lets you map, for example, country subsidiaries with different ranges.

Does the AI also suggest accessories and complementary products?

Yes. Scalera's AI learns from employee behavior, suggests complementary products and accessories, and remembers which products are often offered together.

Is my product data used for other customers?

No. According to Scalera, your completed offers are used only for your own product suggestions; there is no cross-customer AI training.

What is the digital product passport for construction products?

A machine-readable data set introduced by the new EU Construction Products Regulation (EU) 2024/3110. Among other things, it brings together the declaration of performance and conformity across a product's life cycle. The European Commission plans the specific requirements for Q2 2027.

Next steps

• Pick a product line with many requests and go through the checklist.
• Add missing units and conversion factors first; they are quick to maintain.
• Check with your distributors which classification standard they expect.
• Test product matching with a real bill of quantities and your current data.
• 150+ companies already work with Scalera. Book a demo