Original explainer
n8n Merge AI results: match IDs and inspect duplicate keys
Use Matching Fields to join input records and AI results, handling unmatched items, duplicate keys and conflicting fields.
After separate processing, AI results may be ordered differently from original records. Join two synthetic branches by id instead of relying on array position. Both branches must retain stable IDs, and you need access to Merge settings. Documented modes and an offline example were checked; no workflow or AI call was run.
Action and acceptance sequence
Choose a key
Confirm ID names and types in both branches.[1]
Choose output scope
Choose matched rows only or retain every original record.[1]
Inspect duplicates and clashes
Review one-to-many matches and same-name field ownership.[1]
Start with a reordered sample
Original Input 1 contains demo-a and demo-b; Input 2 returns category in the order demo-b, demo-a. Position combines by index rather than ID. This exercise requires matching the same record by ID. If the branches do not retain a common identifier, fix that upstream instead of treating coincidentally equal ordering as identity evidence.[1]
Matching Fields and output scope
Choose Combine, then Matching Fields, and specify id in Fields to Match. Keep Matches behaves like an inner join and retains matches; Enrich Input 1 behaves like a left join and retains Input 1 while adding matching data. Append outputs items from inputs sequentially rather than enriching the same ID. Choose the required behavior, then inspect a deliberately unmatched ID.[1]
Duplicate IDs and name clashes
Multiple Matches can include all matches or keep only the first. Inspect duplicate IDs before choosing, so results are not silently discarded. Input 2 wins name clashes by default; Clash Handling can change priority or add input numbers to field names. Fuzzy Compare is off by default and tolerates some type differences when enabled. A clearer input contract normalizes ID types first.[1]
Acceptance and execution scope
Check ID-to-category ownership, unmatched counts and the source of conflicting values. Stop downstream writes when duplicate keys or unexplained extra outputs appear. Offline checks covered reordered ID matching and duplicate detection. Merge, databases and workflow effects were not executed. Record your actual output mode and duplicate-match policy when adopting the guide.[1]
Article sources
Original editorial content; official descriptions and checking suggestions are distinguished.
Tools in this explainer
Tool directory
n8n
Node-based workflows for connecting services, transforming data, and adding AI steps to business automation.