Writing a search that works
Job titles beat categories, technology names also match skills pulled from the full posting, and common abbreviations are expanded for you.
- Job titles work better than categories. "Frontend Engineer" is matched against posting titles. "Developer" casts wide and ranks worse.
- Technology names are handled differently from role names. Searching "React" or "Kubernetes" also matches the skills we extracted from the full posting, not just the title, because most postings mentioning a technology never name it up top.
- Abbreviations are expanded. gtm, sre, pm, qa, ml, devops, ux and similar resolve to the longer form as well, so you do not have to guess which one the employer wrote.
- Several roles at once. Add more than one and every combination is queried in parallel, then merged into one list.
How the order is decided
Relevance first, recency second. A title that exactly matches what you typed outranks one that merely contains the words, and titles that shift the role, such as manager, director, lecturer, recruiter or intern, are pushed down unless you asked for them. Nothing about that order can be bought.
Where it stops: Search titles are English only. A German-language posting is found by its English title if it has one, and missed if it does not. Native-language keywords have been measured but are not shipped, because a term that returns plausible-looking rubbish is worse than one that returns nothing.
Related
- Searching another country, or everywhereThe country box is a guess from your device on a first visit. Clear it to search everywhere, or type any country or city.
- Sorting resultsFour orders: best match, newest, salary and CV match. Sorting only reorders, it never removes a row.
- When a search returns nothingEither your filters removed everything, in which case the page names the one responsible, or the index holds nothing yet for that query.
Did this answer your question?
Still stuck? Ask us directly, and mention this article so we know where to start.