NER · Named entity recognition
NER (Named Entity Recognition) is the automatic extraction of named entities from a text: people, organizations, places, products. It lets you audit what an engine actually identifies in a page, and therefore whether the main subject comes through.
Standalone definition · 38 words
How does NER work?
A model spots the text segments that designate entities and assigns them a type. Services such as Google’s Natural Language API return entities, their type, their salience and, when possible, a link to a knowledge base.
How do you use it in an audit?
Run a strategic page through an NER tool, compare the extracted entities with the target subject, check that the target entity is detected and salient, and spot parasitic or ambiguous entities. A poorly recognized main entity signals vague text or an ambiguous name.
What are the limits?
Results depend on the model and the language. An entity detected by a tool does not guarantee an engine recognizes it the same way. NER is an editorial diagnostic, not proof of what Google does.
Frequently asked questions
NER and salience: what is the difference?
NER detects entities. Salience measures the importance of each in the text.
Which tools should I use?
Google’s Natural Language API, spaCy or open source models.
Does NER work in French?
Yes, with performance that varies by model.
Related terms
Further reading
- Checklist GEO and LLM / AI 2026
- 2026 GEO Tools Comparison: The Decision Guide
- ← Back to the GEO glossary (114 terms)