The four ingredients
- Text similarity — exact, prefix, word-contains, look-alike spelling, sound-alike, transliteration. Identical scores 1.0; “Nikee” against NIKE scores high on sound.
- Goods overlap — the Nice classes of the mark against the classes you selected. Same class, full weight; no overlap, low.
- Legal status — registered and pending marks count fully; opposed and published slightly less; dead, expired and abandoned marks are shown but heavily discounted.
- Territory — a mark registered where you sell counts fully; elsewhere, less. EU marks cover every EU marketplace; WIPO Madrid designations are expanded to their countries.
The result is 0–100 and comes with reason codes — “identical text”, “sounds alike”, “same goods”, “mark not live”, “outside your territory”, “short mark” — so the number is never a mystery.
The three labels
| Label | Score | Meaning | Do |
|---|---|---|---|
| HIGH REVIEW | 72–100 | Live mark, identical or near-identical, in your goods, in your market | Do not publish as is. Reword, or get advice. |
| REVIEW | 45–71 | Similar text or sound, or same mark in related goods | Ten minutes: open the mark, read the reasons, decide. |
| LOW | 0–44 | Nothing close in the registers we hold, for your goods | Publish; keep an eye on new filings. |
What the score deliberately does not say
It never says SAFE, LEGAL or GUARANTEED. Those words belong to lawyers who have seen the facts. It also never claims coverage it does not have: every result shows the registers searched and the record count of each, and the coverage page shows which ones we have validated record-by-record.
In a listing scan
A listing has many phrases. The listing’s verdict is the highest phrase score, attenuated by how brand-like the phrase is: a hit on your brand field counts in full; a hit on a casual two-word n-gram counts less. Each phrase keeps its own score, so you see which word carries the risk.