Skip to main content
A pre-screen is a knock-out search. It finds earlier marks that contain the same words, not marks that look, sound or mean the same. It does not replace a full clearance search or legal advice.

​
What the API can and cannot find

See Text matching for exactly how words and characters are matched.

​
1. Break the mark into search terms

For a proposed mark LUMEN AIR in class 25:
  • the whole mark: mark_text=LUMEN AIR and mark_text_phrase=lumen air;
  • each distinctive word on its own: mark_text_phrase=lumen; skip descriptive words such as air alone if they would match too much;
  • known variants and translations: lumière, and the mark in the scripts of your target markets.

​
2. Search the relevant registers, classes and statuses

Search the registers of the countries where the mark will be used, including EM for the European Union and WO for international registrations. Restrict to the classes of interest and related classes, and to marks that can still be obstacles:
  • status_category=live,unknown,none keeps registered and pending marks, and also marks whose status is unknown or not stated, so an unmapped status never hides a conflict.
  • In China and Korea, add similar_group_code for the goods you sell; examiners there compare by similar group, not by class.
  • International registrations designate countries: add designated_country to the WO search to keep only those covering your markets.

​
3. Review the hits

Summaries carry the mark, status, dates, primary owner and Nice class count. Retrieve the candidates that need a closer look, with the specification:
goods_services gives the specification per class, with a translation where one exists. Check status.office_status and the dates: a registered mark past its expiry date and grace period may be vulnerable, and a pending application may still be refused.

​
4. Search figurative elements

Marks with no words, or whose words differ from yours, can still conflict through a device. If your mark has a distinctive device, search its Vienna codes in the same scope:
A one- or two-part Vienna prefix must be combined with jurisdiction, nice_class or primary_owner_country.

​
When a term matches too much

A very common word or a single Chinese character can match more than 100,000 records and get 422 query-too-broad. Narrow it with fewer registers, Nice classes or a filing-date range. If the word is that common, it is probably not the distinctive part of the mark.

​
Recording what you searched

For your file, record the date, the request_id of each search, the parameters, and the freshness of each register from GET /v1/jurisdictions (last_checked_at). Results reflect the registers as Pinakel last loaded them, not the office’s live state.