Title: How I Built an Automated Data-Cleaning Pipeline with n8n (So I Never Touch Messy Spreadsheets Again)

If you've ever collected customer data through a Google Form, you already know what happens next: the spreadsheet turns into chaos. Names in different cases, duplicate emails, phone numbers missing digits, dates written five different ways in the same column.

I got tired of manually cleaning this every week, so I built an automated pipeline in n8n that fixes all of it the moment a form is submitted.

The Problem

A raw form submissions sheet usually looks something like this:

Full Name	Email	Phone	City	Registration Date
John doe	[e-mail]	1712345678	Dhaka	01/05/2025
JOHN DOE	[e-mail]	1712345678	Chittagong	May 1, 2025
tanvir ahmed	[e-mail]	171122	Dhaka	2025/06/12

Same person entered twice with different casing. A copy-pasted email attached to the wrong name. A phone number missing half its digits. Three different date formats in three rows. Multiply this by a few hundred submissions and manual cleanup becomes a part-time job.

The Solution: An n8n Pipeline

Here's the flow:

flowchart LR
    A[Form Submission Trigger] --> B[Messy Data]
    B --> C[Filter Code]
    C --> D[Clean File]
    C --> E[Summary Report Filter Code]
    E --> F[Summary Report]

A form submission triggers the workflow, pulls in the raw row, runs it through a Code node that handles all the normalization logic, then splits into two outputs — a clean master file and an auto-generated summary report.

What the Cleaning Logic Actually Handles

• Names → consistent title-case formatting
• Emails → deduplicated and validated against the right record
• Phone numbers → normalized into one consistent format (in my case, Bangladeshi numbers into 01XXXXXXXXX)
• Dates → every format converted into YYYY-MM-DD
• Customer ID → auto-generated and assigned per record

The trickiest part was date and phone parsing — real-world form input is way messier than test data. A rough example of the kind of normalization logic involved:

function normalizePhone(raw) {
  const digits = raw.replace(/\D/g, '');
  if (digits.length === 10 && !digits.startsWith('0')) {
    return '0' + digits;
  }
  return digits;
}

(Simplified for illustration — the actual node handles a few more edge cases, like missing digits and stray country-code prefixes.)

The Result

Customer ID	Full Name	Email	Phone	City	Registration Date
CUST-0001	John Doe	[e-mail]	01712345678	Dhaka	2025-01-05
CUST-0002	Sarah Khan	[e-mail]	01987654321	Chittagong	2025-05-03
CUST-0004	Tanvir Ahmed	[e-mail]	01711223344	Dhaka	2025-06-12

Same data, but now every row is consistent, deduplicated, and immediately usable — in a CRM, a mailing list, or anywhere else it needs to go.

Stack

• n8n for orchestration
• Google Forms & Sheets for data collection
• Excel as the cleaned output

Why This Is Worth Automating

• No more weekly manual cleanup
• Fewer errors from duplicate or mistyped records
• A summary report that's ready the second new data comes in
• Scales identically whether you get 10 submissions a week or 10,000

If you're building something similar, or you've handled messier real-world data than this, I'd genuinely like to hear how you approached deduplication or date parsing — always looking to improve the edge-case handling.

Contact With me Linkedin : www.linkedin.com/in/bastullah-batas-54ab23335