Back to the guides

PRACTICAL SOFTWARE GUIDE

n8n Merge node: reconcile matching fields, position and unpaired items

Merge by a stable business key when records can arrive in different orders, and state what happens to unmatched items before activation.

SOURCE-BASED ANALYSIS · NO HANDS-ON RATING

Choose the relationship, not the most convenient mode

n8n's Merge node can append streams, combine items by matching fields, combine by position, create all possible combinations or choose one branch. Append keeps items sequentially without joining their fields. Matching fields behaves like a configurable join, while position assumes the items in each input are aligned by order.

Build two controlled inputs

Create synthetic customer and order streams with stable customer IDs. Include one matched pair, one customer without an order, one order without a customer and the same records in a different order. Predict the result before choosing a mode.

FixtureMatching fieldsPosition
Same order in both inputsCorrect pairAppears correct
Different input orderCorrect pair by IDWrong pair risk
Unmatched customerKeep or exclude by output typeDepends on index and option
Conflicting field nameFollow clash policyFollow clash policy

Declare unmatched and clash behavior

Matching Fields can keep matches, non-matches, everything or enrich one input with the other. Fuzzy comparison can tolerate some type differences, but it may also conceal inconsistent source typing. Field-clash handling determines which value survives when both streams use the same name.

  1. Name the real relationship between the streams.
  2. Select a stable key and normalize its type.
  3. Run matched, unmatched and reordered fixtures.
  4. Choose the required join-style output.
  5. Review field-name clashes and nested data.
  6. Count expected and actual output items.
  7. Reject positional joining when order is not guaranteed.

Watch multiplicative output

All Possible Combinations creates every input pairing and can multiply output dramatically. Use it only when that Cartesian relationship is intended. Accept the merge when each output item traces to named inputs, unmatched records follow the written policy and reordering does not silently change meaning. This guide does not claim a live dataset result.

Sources and verification

Official documentation checked on 2026-10-29. Worked examples are hypothetical planning exercises, not measured product benchmarks.

Continue your evaluation

Practical guides for this topic

Put your tools to work.

Need help choosing or setting up your tools? Tell Koldia what you want to improve.

Discuss your project