Choosing a search step

A POST /search request is a pipeline of search steps, each specialized for one database. Picking the right step (or combination of steps) depends on what you’re looking for.

Full text or explorative search → Elasticsearch query

Use the Elasticsearch query step when you’re searching by content: free text with fuzziness, exact keyword matches, numeric or date ranges, and similar document-level lookups. It only ever returns a flat list of elements - it has no notion of relations between them.
See writing an Elasticsearch query.

Structural, relation-aware search → Cypher path query

Use the Cypher path query step when the question is about how elements relate to each other, e.g. “all comments on blog post X” or “all plants tagged blue”. It queries the graph directly and returns paths, which is the only step able to reflect relations in the result.
See writing a Cypher query.

Converting ids to full elements → Element hydration

Both steps above only return element ids (plus a bit of metadata like Elasticsearch’s score). Add an Element hydration step - typically last in the pipeline - to turn those ids into a full collection response with all element properties included.

Combining steps

Steps run in order and can pass their results on to later steps, so you’re not limited to a single database per request. A common pattern is to narrow candidates structurally with a Cypher step, then re-rank or filter that set by content with an Elasticsearch step (or the other way around), before hydrating the final result.
See parameters and expressions for how results flow between steps, and the reference examples for complete request/response pairs of several such combinations.