Fix: n8n workflow data not reaching the output
A green n8n execution only proves that nodes completed without throwing an error. Trace every connection and inspect every item to find branches that fetch data but never reach the output.
Last updated: 2026-06-08 · Tested on n8n 1.x · Approximately 4 min read
Symptoms — what you're seeing
- The n8n workflow runs on schedule and shows a successful execution, but the final output contains only some of the records you expected
- Several HTTP Request, RSS Feed Read, database, or API source nodes return data, but that data never appears in the scoring, filtering, notification, or storage step
- The workflow has no visible errors, failed nodes, or red execution indicators
- A paid API is being called on every run even though none of its records reach the final result
- The workflow was built with many source branches and nobody has recently traced each branch from its fetch node to the final output
Why this happens
n8n reports node execution status, not business correctness. An HTTP Request node can return a valid response, show a green checkmark, and still be irrelevant to the final result if its output is not connected to the next node in the data path.
The usual failure is a disconnected branch. The source node fetches records, but the connection stops before the Merge node, Code node, AI scoring step, database insert, or notification node. Nothing crashes because n8n has no reason to treat an unused output as an error.
The same problem appears when a workflow is edited after it was working. A source gets duplicated, a new branch is added beside the existing ones, or the combine step is replaced. The source still runs because it remains connected to the trigger, but only the original branches feed the output.
Do not confuse this with a source returning zero items. A zero-item source is a data or API problem. A source returning items that never appear in the combine node is a topology problem. You can have both at the same time, so inspect the execution data for every branch.
If one of the source nodes is failing with an authentication error instead, use the diagnostic steps in the n8n HTTP 401 unauthorized fix. That is a request failure, not an unconnected-output failure.
Stuck on this for more than 30 minutes? Book a 30-min scope call Email Basit. I'll get on Zoom, look at your workflow, and either fix it on the call or scope what it'll take. No upsell, no decks.
The fix, step by step
- Start with the final output node. Open the last node that produces the result you care about: the notification, database write, AI scoring output, or response. Work backward one connection at a time. Do not start with the trigger or trust the visual grouping of nodes on the canvas.
- List every source branch. Write down each node that is supposed to contribute data. Include HTTP Request nodes, RSS Feed Read nodes, database queries, sub-workflows, and manually added branches. In a job-search workflow, this might be one source per API or feed.
- Trace each source to the combine step. Follow the actual lines on the canvas. Every source that is meant to contribute must have a continuous connection into the Merge or other combine node. A node sitting beside the path is not connected. A node that runs successfully is not necessarily part of the output.
- Inspect the execution data at the combine node. Open a recent execution, select the Merge or combine node, and inspect its input data. Confirm that records from each expected source are present. Check the item count and a source-identifying field such as
source,feed, orprovider. If a source returns items upstream but none appear at the combine node, the missing connection is the cause. - Check the combine node configuration. Confirm that the Merge node is using the intended input structure and mode for the workflow. The wiring and the mode must agree. If the workflow expects an append-style result, configure the node to preserve items from each connected input instead of matching records by a key that does not exist.
- Run a controlled test with distinct source labels. Temporarily add or preserve a source label on each branch before the combine step. Run the workflow once with a small test range, then count the labels at the output. This makes it clear whether the problem is a missing connection, a filter, or a combine configuration.
- Validate all incoming items in a Code node. If you need to audit the output, process the complete input collection. Do not use only the first item with
$input.first(); that can make a multi-source workflow appear valid while ignoring the rest of the records.// Code node: Run Once for All Items const items = $input.all(); const counts = {}; for (const item of items) { const source = item.json.source ?? item.json.provider ?? "unknown"; counts[source] = (counts[source] ?? 0) + 1; } return items.map((item) => ({ ...item, json: { ...item.json, _audit: { source: item.json.source ?? item.json.provider ?? "unknown", totalItemsAtCombine: items.length, itemsBySource: counts, }, }, }));This example keeps every incoming item and adds an audit object to each one. Replace
sourceandproviderwith the field your source branches actually set. If the total count is correct but one expected source is absent fromitemsBySource, inspect the branch connection and the node immediately before the Code node. - Remove or disable unused paid branches after verification. Do not cancel an API because one execution was empty. First confirm that the branch is either intentionally unused or duplicated by another source. Then disable the HTTP Request node, remove its credentials if they are no longer needed, and cancel the subscription outside n8n.
Common variations
- Only the first source reaches the AI node. The combine step is receiving one input, or a Code node is returning only
$input.first(). Inspect the node input count and process$input.all()when the workflow is designed to handle multiple items. - All sources reach Merge, but some records disappear afterward. The problem is downstream. Check IF filters, Remove Duplicates, Item Lists operations, Code node filtering, and any AI scoring threshold after the combine step.
- The branch runs but contributes zero items. Inspect the HTTP response, pagination, date filters, and response mapping. A successful HTTP status does not mean the response contains records your next node can process.
- The workflow was cloned from an older version. Compare the current canvas with the last known working execution. Cloning and rearranging nodes can leave valid source nodes outside the active data path.
- The output count changes between scheduled runs. Check source availability, pagination, deduplication keys, and time windows. A topology issue is consistent; a changing count usually adds a source-data or filtering issue.
When this isn't the issue
If every expected source appears at the combine node, the wiring is not the missing piece. Continue down these forks:
- The schedule runs at the wrong time. Check the workflow and instance timezone, daylight-saving behavior, and the Schedule Trigger configuration. The n8n schedule trigger timezone fix covers that branch.
- The AI scoring step fails or processes only part of the batch. Inspect the item count before and after the AI node, then check for OpenAI 429 responses, batch size, and wait behavior. Use the n8n OpenAI 429 rate-limit diagnostic when executions stop at the model request.
- The combine node receives all data but the final action is incomplete. Inspect the final filter, database constraint, API response, or notification payload. A successful upstream path does not prove that the downstream system accepted every item.
If the workflow is still paying for sources that do not contribute, disable the suspected branch for one controlled run and compare the final output. Keep the execution data from before and after the change. The goal is to prove which source changes the result, not to infer it from a green status icon.
Stuck on this for more than 30 minutes? Book a 30-min scope call Email Basit. I'll get on Zoom, look at your workflow, and either fix it on the call or scope what it'll take. No upsell, no decks.
Related pages
Need this workflow customized or debugged? Book a 30-min scope call →
Stuck configuring this workflow?
I build and fix n8n workflows for agencies and SaaS teams. 30-minute scope call — bring the problem, leave with an estimate.