Anyone preparing private-label matching faces a typical problem: the same products, different names.
Internally your disposable gloves are called “White disposable gloves size M, pack of 100”. In a competitor shop they appear as “Disposable gloves white M box/100”. No shared code, no shared name — still the same product.
Pure text matching does not solve this reliably. You need clean product data, clear domain rules and a short acceptance process. The AI does not decide in a vacuum; it follows your logic.
The more structured your data and the clearer your rules, the faster we reach reliable match quality.
Wondering why private-label matching matters in the first place? We explain it in this blog article:
Same product, different name. Your pricing strategy is off.
The 5 building blocks for successful private-label matching
1. Product data: what you need to provide
We do not match private labels by name alone, but by the product attributes that actually matter. For each item we need:
- Unique article number per variant and pack size (SKU)
- Product name — as used internally
- Category path — e.g. “PPE > Gloves > Disposable gloves”
- Structured attributes: dimensions, material, colour, quality, standards
- Pack unit / pack size — e.g. pack of 100, box of 10
The more complete these fields are, the faster we can start.
2. Matching rules: what counts and what does not
Not every attribute is equally important in every category. You need to define this once:
- Allowed category mappings: Which external categories may be compared with yours?
- Tolerances per attribute: May a pack of 95 count as a match for a pack of 100? From which deviation onwards should it not?
- Hard exclusions: Printed vs. unprinted, sterile vs. non-sterile: such differences must never be matched, however similar the rest is.
- Attribute priorities: What matters more: material or standard? Colour or size?
These rules sound like extra work, but they only need defining once — and they make the difference between usable and reliable matching.
3. Data quality: what “clean” actually means
Good rules are useless if the data is inconsistent. Three common problems:
- Inconsistent units: “100 pieces”, “100 pcs”, “PU 100” — the same thing, but a system sees them as different.
- Conflicting duplicates: The same product twice in the system, with slightly different attributes.
- Empty required fields: If, for example, the pack unit is missing, pack-size logic cannot work.
A short data check before the project starts saves time.
4. Pack-size logic: mapping quantities cleanly
Many private labels come in several pack sizes. A pack of 50 and a pack of 100 are not the same product, but they still need to be made comparable.
That requires a clear rule: how should quantity variants be mapped? Should we normalise to unit price? Are there quantity bands that must never be compared?
This is not a technical question — it is a business decision you need to take once.
5. Quality assurance: how we refine together
No matching is perfect from day one. That is why we work iteratively:
- Gold set: You provide 20–50 examples with a known outcome: “these two products are a match”, “these are not”. That is the benchmark for the first calibration.
- Thresholds: Together we define from which similarity score a match is accepted automatically, and when a manual review is triggered.
- Iterative refinement: After the first results we adjust rules and tolerances until the quality is right.
Checklist: what we need before project start
| What | Details |
| Unique SKU | Per variant and pack size |
| Category path | Per product, structured |
| Attributes | Dimensions, material, colour, quality, standards |
| Pack unit | Pack size per SKU |
| Hard exclusions | printed/unprinted, sterile/non-sterile |
| Tolerances | Per attribute and category |
| Gold set | 20–50 reference examples right/wrong |
Private-label matching is not a black box.
With the right data and clear rules it becomes a transparent, reliable process.
In our blog article “Private-label matching pricing strategy” we explain why it is the prerequisite for your pricing strategy to work in the most important part of your range.
Talk to us. We will look together at where you stand today.
