How I Set Up Nightly GA4 Data Pulls into Google Sheets

The Problem: Data Retention & Agility

Most marketers don't know that detailed user-level data in standard GA4 properties expires after 2 months or 14 months (depending on your settings). If you want to analyze a cohort from 2 years ago, you are often out of luck.

I didn't want my data to have an expiration date. I also hated the latency—sometimes waiting 24-48 hours for data to settle in the interface.

So, I built an automated "Nightly Data Pull" system. Think of it as a methodical data harvest.

The Engineering: How It Works

This isn't enterprise-grade rocket science; it's a scrappy, robust script running on Google Apps Script.

Step 1: The 'Yesterday' Trigger

Every morning at 3:00 AM, my script wakes up. It calculates yesterday's date (today - 1). This ensures I'm always asking for a complete 24-hour cycle.

Step 2: The Dimensions & Metrics

I don't pull everything. I pull 3 specific tables to different sheets:

1. Traffic_Daily: date, sessionSourceMedium, campaignName, sessions, activeUsers.
2. Pages_Daily: date, pagePath, views, entrances.
3. Events_Daily: date, eventName, eventCount, conversions.

Step 3: Stacking the Data

The script doesn't overwrite the sheet. It appends rows. It finds the last empty row and pastes the new data. Over time, this builds a massive, vertical database inside the sheet.

Issues Faced (The "Gotchas")

1. Sampling on "Day" Dimension

Surprisingly, even daily pulls can get sampled if the property is huge. I solved this by splitting the request. If I pull "Traffic" separate from "Pages," sampling is rare. If I try to combine "Source" + "Page" + "Device" + "City" in one request, sampling hits.

2. The 10-Million Cell Limit

Google Sheets has a hard limit of 10 million cells. My "Pages_Daily" tab hit this after 18 months for a large client.

The Fix: I implemented a "Yearly Archival" logic. The script checks the date. If it's January 1st, it creates a new spreadsheet file "GA4_Data_2025" and starts writing there.

Why Go Through This Trouble?

1. Historical Analysis That Actually Works
I can now run a query: "Show me the performance of the 'Black Friday' campaign for the last 3 years."
In GA4, this is difficult because campaign names might have changed, or data expired. In my Sheet, it's just a SUMIFS formula.

2. Blending with Non-Google Data
I have another script that pulls Shopify sales data nightly. Because both datasets live in Sheets and share a date key, I can blend them perfectly. I can calculate "Total Profit / GA4 Session"—a metric impossible to get in GA4 alone.

3. Owning the Asset
Data is an asset. If Google Analytics goes down, or changes its pricing (like the UA sunset), I have my CSVs. I am platform-independent.