n8n Marketing Attribution: Tie Revenue Back to Channels

Marketing teams run on a spreadsheet that's always slightly wrong. Spend lives in one tab, leads in another, closed revenue in the CRM, and once a week somebody reconciles them by hand to answer one question: which channel actually made money? n8n marketing attribution automates that reconciliation, runs it daily instead of weekly, and flags the channel bleeding budget before the month closes.

The hard part isn't the arithmetic. Cost per acquisition is spend divided by leads; ROI is a one-line formula. The hard part is keeping the data joined: matching a closed deal back to the UTM source that originated it, three weeks and four touchpoints later. That join is where manual attribution breaks, and it's exactly what a workflow does without complaint.

This covers wiring spend and revenue together, computing CAC and ROI per channel, picking an attribution model honestly, and pushing the result somewhere people will actually see it.

What you can automate

• Pull spend per channel from your ad exports or a Sheet
• Join closed deals back to their originating channel by UTM source
• Calculate cost per acquisition and ROI for every channel daily
• Flag any channel that breaches a CAC threshold before spend compounds
• Append a timestamped snapshot so you can track channel trends over time
• Send a daily summary to Slack or email instead of digging through tabs
• Handle channels with zero leads as data gaps, not silent failures

What the existing attribution templates get wrong

The most-shared option is the revenue attribution template on n8n.io (Automate Marketing Revenue Attribution & ROI Analytics with Airtable and Slack). It runs nightly, matches closed deals to lead sources in Airtable, computes ROI as ((Revenue - Cost) / Cost) × 100, and posts a report to Slack. Genuinely useful. But its attribution is source-based only, pure first-touch. Its own notes flag no multi-touch, no time-decay, no anomaly detection, and no CAC payback analysis. So it answers "which source got credit" but not "which channel is getting expensive fast."

The BANT-style qualification flows that often come up in the same searches (Automate Lead Qualification & Multi-Channel Follow-up with AI) don't touch attribution at all; they score and route individual leads, with no channel-level spend or ROI math. The gap a sharper post fills is the part both skip: the daily CAC-per-channel calculation with a threshold alert, so you catch a channel breaking its budget while you can still cut it.

Here's the stance worth stating plainly: most teams obsess over the attribution model and ignore the threshold. Whether you run first-touch or time-decay matters far less than whether anyone notices when a channel's CAC doubles. Pick the simplest model your data supports, ship it, and spend your energy on the alert that fires when a channel goes underwater. The model is a debate; the alert pays rent.

The Attribution Pipeline

Trigger (daily schedule)
  → Process (join spend + leads, compute CAC/ROI)
    → Route (within threshold / breached)
      → Action (snapshot to dashboard + Slack alert)

1. Collect spend and outcomes

A n8n-nodes-base.scheduleTrigger (v1.2) fires each morning. It reads two sources via n8n-nodes-base.googleSheets: a spend sheet (channel plus spend) and a leads or deals sheet (channel plus qualified leads or closed revenue). For teams already in a CRM, swap the leads read for an n8n-nodes-base.httpRequest (v4.2) call to HubSpot or Salesforce. Keep UTM source as the join key on both sides; that consistency is the whole game.

2. Process the math

A n8n-nodes-base.code node (v2, jsCode) does the work code is good at and an LLM has no business touching: group spend by channel, group leads by channel, then compute CAC as spend ÷ leads and ROI from revenue per channel. This is where the attribution model lives. First-touch and last-touch are simple lookups. Multi-touch, whether linear or time-decay, weights the touchpoint history right here in JavaScript. Make the model an explicit, commented choice in the node, not an accident of how the data happened to sort.

One failure mode to handle: a channel with spend but zero leads. Naive division gives you a divide-by-zero or an infinite CAC that poisons the summary. Catch it and label it a data gap so the run doesn't fail and the report stays readable.

3. Route by threshold

A n8n-nodes-base.if node checks each channel's CAC against a configured ceiling. Within budget, log and move on. Over the ceiling, branch to the alert path. The threshold is the single most useful number in the workflow, and it belongs in a config field you can edit without opening the Code node.

4. Act: snapshot and alert

Every run appends one row per channel to a dashboard sheet via n8n-nodes-base.googleSheets, building the trend line that tells you whether a channel is drifting or steady. When a channel breaches the ceiling, a n8n-nodes-base.slack message names the offending channels and their CAC, so the marketing lead sees it with morning coffee instead of at the month-end review when the budget's already spent.

5. Keep the join clean

None of this works if the UTM data is messy. A periodic check that UTM sources match your taxonomy, and that parameters haven't been stripped by a redirect, keeps the attribution honest. A n8n-nodes-base.code node can validate incoming sources against an allowed list and quarantine the unknowns for review rather than silently miscrediting them.

Implementation patterns

Pattern 1, model-explicit Code node. Put the attribution choice in one commented place. Don't let first-touch happen by default because that's how the rows were ordered.

Code (v2):
  // ATTRIBUTION MODEL: last-touch
  // switch to time-decay by weighting touchpoints
  const cac = totalSpend / Math.max(leads, 1);

Pattern 2, zero-lead guard. Treat any channel with spend and no leads as a flagged data gap, not a math error. Keeps the daily summary clean and surfaces tracking breakage early.

Pattern 3, threshold-first alerting. Lead with the breach alert, not the full report. A daily firehose gets muted; an alert that only fires when a channel goes underwater stays read.

n8n nodes you'll use most

Getting started

1. Keep a spend sheet (channel, spend) and a leads sheet (channel, leads), joined on UTM source.
2. Add a Schedule Trigger set to fire each morning.
3. Read both sheets, then add a Code node to group by channel and compute CAC and ROI.
4. Make the attribution model an explicit, commented choice in that node.
5. Guard against zero-lead channels so the run doesn't break.
6. Append a snapshot row per channel to a dashboard sheet for trend tracking.
7. Add an IF on the CAC threshold and a Slack alert for any breach.

For teams that want the cost side broken out on its own, n8n CRM automation covers keeping the deal and source data clean enough that the join actually holds.

Attribution tells you where the good leads come from; the next step is acting on it faster. The scoring side in n8n lead scoring automation shows how to weight inbound from your best-performing channels, and the Ad Performance Optimizer template closes the loop by pulling the daily ad metrics that feed the spend side of this calculation. Know your channel ROI before the budget meeting, not after.