This is the documentation that ships inside the ZIP — the same README, the same start-here steps, the same troubleshooting page written from our own failed runs. Generated from the shipping files (sha256 9f5a9f816769) by make-docs.mjs, not retyped; the site build fails if these files change and this page does not.
On this page
LeadFlow Automation Pack (n8n)START HERE — LeadFlow Automation PackDEMO.md — 2-minute walkthrough of LeadFlowINSTALL — running n8n for this packTROUBLESHOOTING — LeadFlow Automation PackFour ready-to-import n8n workflows for B2B lead generation — compliant by design (no LinkedIn scraping, no account bans) — plus a zero-setup demo that runs with no credentials.
| File | Purpose |
|---|---|
| workflows/00-quick-demo.json | Zero-setup demo: 6 messy rows → clean leads, patterns, drafts (no credentials, no network) |
| workflows/01-lead-import-csv.json | CSV (your export, as-is) → normalize headers → dedupe by email → Google Sheets |
| workflows/02-webhook-lead-capture.json | Webhook (any form) → Google Sheets |
| workflows/03-ai-email-drafter.json | Leads → AI-personalized email drafts (DeepSeek) → Google Sheets |
| workflows/04-email-patterns.json | Leads → generate candidate email patterns (FREE, no API) → Google Sheets |
workflows/00-quick-demo.json and Execute it — no credentials, nothing written. Then configure the rest.Sheet1 (your leads), Drafts (03 writes here), Email Patterns (04 writes here). Output tabs are separate on purpose — re-running a workflow can never treat generated rows as fresh leads.YOUR_GOOGLE_SHEET_ID) with your file, pick the tab listed above, then choose your Google Sheets credential.Load CSV node is preset to /home/node/.n8n-files/sample-data/leads-sample.csv, which the shipped Docker compose (setup/) mounts read-only - run it as imported. On npm or your own compose, n8n 2.x restricts file access to ~/.n8n-files (N8N_RESTRICT_FILE_ACCESS_TO): copy the CSV there or widen the variable, then point the node at your path. On n8n Cloud, file reads are limited to /home/node, so run 01 self-hosted and start Cloud users on 02/03/04.It renames the columns it recognises, so a Pipedrive / HubSpot / Mailchimp export runs without hand-editing the file:
email, Email, Email Address, Email - Work, Work Email, Business Email, Email 1First Name + Last Name (or First/Last), or one Name / Full Name cell — Petrović, Marko is read first-name-first, Acme, Inc. is left aloneCompany, Company Name, Organization, AccountTopic, Note, Notes, Reason (workflow 03 uses it as the draft subject)Every other column (Phone, Owner, …) is appended to your sheet untouched. If no address column can be found, 01 stops with No email column found. Headers seen: … instead of writing rows with an empty email; a column called Email Status is deliberately not treated as an address.
Drafts, Email Patterns) and skip rows that are already drafted/enriched, so re-runs are safe and idempotent.Email - Work, Organization), HubSpot (Work Email, Company Name), Mailchimp (Email 1) and a single Petrović, Marko cell all produce the same clean rows as the plain email/firstName file, and the same duplicate person collapses to one row (checked on every build by test-wf01-headers.js).verify-pack.ps1; workflow 01 uses readWriteFile + extractFromFile, the current replacements for the deprecated Read Binary Files / Spreadsheet File nodes).4 importable n8n workflows for compliant B2B lead generation (no LinkedIn scraping, no account bans), plus a zero-setup demo that proves your n8n runs the pack before you connect anything.
workflows/00-quick-demo.json.Start - click Execute workflow.Summary - what just happened: 6 messy sample rows -> 3 clean leads, 15 email patterns, 3 drafted emails, 0 emails sent, 0 network calls, no credentials used.It runs entirely in Code nodes: it reads no file, writes nothing, calls nothing. If this works, your n8n can run the pack. The other four workflows add Google Sheets as storage.
setup/INSTALL.md) or use n8n Cloud.01–04) in workflows/.Sheet1: firstName, lastName, company, domain, email, topic. (Your CSV does not have to use those names — workflow 01 renames Email - Work, Work Email, First Name, Company Name and similar on the way in.) Add two more empty tabs: Drafts (workflow 03 output) and Email Patterns (workflow 04 output).Authorization, Value = Bearer <your DeepSeek key>./home/node/.n8n-files/sample-data/leads-sample.csv, and the shipped Docker compose (setup/docker-compose.yml) mounts the pack's sample-data/ exactly there, read-only — so the Docker route runs it as-is: set your sheet, Execute, expect deduped rows appended.~/.n8n-files (env N8N_RESTRICT_FILE_ACCESS_TO). Copy sample-data/leads-sample.csv into ~/.n8n-files/ and point the node at that absolute path, or widen the env var to the folder you keep it in./home/node, so run 01 self-hosted (Docker/npm) - on Cloud, start from 02/03/04 with your leads already in the sheet.README.md (overview) · DEMO.md (2-min walkthrough) · setup/INSTALL.md (n8n install) · TROUBLESHOOTING.md (sheet, credentials, 401, CSV path, webhook, patterns).LICENSE.txt.You need a working n8n instance. Any of these routes works; pick one.
../START-HERE.md).npm install -g n8nn8n startNotes: default DB is SQLite in ~/.n8n. Requires Node.js 20+.
setup/ folder: copy .env.example to .env and set N8N_ENCRYPTION_KEY to any long random string (keep it — it decrypts your stored credentials).docker compose up -dOptional in .env: GENERIC_TIMEZONE (default UTC).
The compose file already mounts this pack's sample-data/ folder read-only at /home/node/.n8n-files/sample-data - inside the folder n8n 2.x allows file nodes to read by default (N8N_RESTRICT_FILE_ACCESS_TO defaults to ~/.n8n-files). So workflow 01 reads its preset path without extra setup or env changes. Keep the compose file in setup/ (the mount path is relative to it). For Route B (npm), copy sample-data/leads-sample.csv into ~/.n8n-files/ instead.
../START-HERE.md for the 15-minute pack setup.../TROUBLESHOOTING.md.Every item below was hit (and fixed) during our own end-to-end testing. Check here before emailing support.
Sheet1 on English accounts, Лист1/Hoja1 elsewhere. A mismatch fails silently with 0 rows written.YOUR_GOOGLE_SHEET_ID and no credential selected.Authorization, Value = Bearer sk-... (the word "Bearer" plus one space, then the key). A missing "Bearer" prefix also yields 401.Two different causes, same node - check which error you get.
"Access to the file is not allowed." (n8n 2.x, the default)
~/.n8n-files unless you change N8N_RESTRICT_FILE_ACCESS_TO. That is why our sample CSV is mounted at /home/node/.n8n-files/sample-data/, not at /data/.setup/docker-compose.yml): nothing to do - the folder is mounted there read-only and the node is preset to it.- ./sample-data:/home/node/.n8n-files/sample-data:ro).~/.n8n-files/, or set N8N_RESTRICT_FILE_ACCESS_TO to the folder holding it, then point the node at that absolute path./home/node). Run workflow 01 on self-hosted n8n, or skip the CSV step on Cloud and start from 02/03/04 with your leads already in the sheet."ENOENT: no such file or directory"
C:/Users/you/leads.csv).Email - Work, Work Email, First Name, Company Name and similar to the columns it uses (see §10). If it finds no address column it says so instead of writing blank rows.typeVersion of some shipped nodes (Google Sheets 4.5, Read/Write Files 1.1, Extract from File 1.2)./webhook-test/..., production URL is /webhook/... - use the right one for the mode.first.last@, firstlast@, f.last@, flast@, first@). They are unverified.Drafts, Email Patterns) and skips rows that already carry a draft or pattern field. If you imported an older build, re-import 03/04 and re-point the append node at its own tab.Yes, no editing needed. 01 renames the columns it recognises to the ones the rest of the workflow uses:
email, Email Address, Email - Work, Work Email, Business Email, Email 1, E-mailFirst Name + Last Name, or one Name / Full Name cell (Petrović, Marko is reordered; Acme, Inc. is left as it is)Company, Company Name, Organization, AccountTopic, Note, Notes, ReasonEverything else (Phone, Owner, custom fields) is appended to the sheet as-is, so nothing is lost. If the run stops with No email column found. Headers seen: …, the list it prints is your file's real header row: rename the address column to email (or one of the names above) and run again. A column named Email Status / Email Valid is deliberately not used as an address.
The characters were destroyed before workflow 01 saw them, or they were written by Excel in an encoding nobody asked for. The two shapes have different answers.
Shape A — the cell holds ???? or ?or?e. Excel on a Serbian, Polish, Czech or Hungarian Windows writes CSV in Windows-1250 by default. n8n's Extract from File node decodes every CSV as UTF-8, so each byte above 0x7F that is not valid UTF-8 becomes a replacement character and the name is gone at that point - not lost later in the sheet.
Workflow 01 therefore carries a Decode CSV bytes node ahead of the parser. It reads the raw bytes and picks the first encoding that is provably correct, in this order:
The node rewrites the file as UTF-8 and records the choice in encoding_used, so the CSV parser downstream only ever sees valid UTF-8. Check that field on the node's output if you want to know which route your file took. It does not guess between 1250 and 1252 by "most accented letters": a wrong guess silently renames people.
If the cell already shows ? after the Decode node, the bytes were replaced before the file reached n8n - by a previous save in a text editor, by a mail gateway, or by an export that was already damaged. No decoder can restore a byte that was replaced; the honest fix is to re-export from the source system. This is the one failure the pack does not fix, and it is stated in README's Limitations on purpose.
Shape B — the cell holds François or KováÄ. That is valid UTF-8 that some earlier tool decoded with a single-byte encoding and saved again. Because the text *is* valid UTF-8, the Decode node correctly leaves it alone. The free browser tool at the LeadFlow site reverses this shape cell by cell (it rebuilds the bytes and checks that the result is strict UTF-8), a file at a time; the pack's workflow 01 does not, so repair those rows before import or hand the file over as described on the site.
UTF-16 exports. Excel's "Unicode Text (*.txt)" and some CRM exports are UTF-16. Extract from File decodes UTF-8 only, so those arrive full of null bytes and the header is unrecognised. Save as CSV UTF-8 instead, or convert once with your spreadsheet's "Save As -> CSV UTF-8 (Comma delimited)".
Reply to your purchase email with: n8n version, the failing workflow number, and the exact error text. Refund window and terms are in LICENSE.txt.
What this page is not: it is not the product. The ZIP contains the four importable workflow JSON files, the zero-setup demo workflow, the sample CSV, a Docker compose that mounts it correctly, the commercial-use license — and these documents. See what is in the pack →
Honest limits, repeated here: email patterns are candidates, not verified addresses, and you confirm each one before sending. Workflow 03 writes drafts to a sheet; it never sends email. n8n must be installed, and Google Sheets needs your own Google account.