How do I export attribution paths from GA4?

In GA4, open Advertising, then Attribution paths. Pick only your purchase key event and the longest date range you trust, then click Share this report, Download File and Download CSV. If the report has filters or the wrong key events selected, the file inherits them, so check those before you click.

What one store's data shows

One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.

Read it as a long tail. In one store's export, just 14 distinct one-touch paths hold 79.5% of revenue. The other 3,656 distinct paths, every journey with two or more touches, share 20.6% on the Journeys sheet. These are distinct sequences of channels, not purchases: the file has no order counts at all.

The slow end is the interesting end. On the Journeys sheet, journeys of ten or more touches took 16.0 days to buy. That is the part of the file a screenshot rarely shows. GA4 sorts the table by the paths with the most conversions, so the first screen is all head and no tail. A screenshot of the top paths is a postcard, not the trip.

One trap for anyone who works with the file: the touched view on the Channels sheet adds up to 110.4% by design. A journey that touched two channels counts once for each, so never add touched shares and call the total revenue.

Row by row, the file does not show cause. A path lists the channels a buyer touched, in order, not the one that changed their mind. And with no spend, margins or refunds in the export, it cannot give you ROAS or profit either.

Why does the quick answer go wrong?

The quick answer is right about the buttons and silent about the defaults. By default, GA4 selects every key event in this report and adds them together. Your export then mixes purchases with sign-ups, form fills and whatever someone marked as a key event years ago.

The chart above the table uses the Paid and organic channels last click model unless you switch it. The table splits paths by Primary channel group unless you pick Source, Medium or Campaign instead. And the report only displays paths up to 20 touchpoints long. None of that is wrong. It just means the file answers the question you set up, not the one you meant.

Short date ranges bring their own trouble. Google may withhold data from a report with a narrow date range when user or event counts are low. A longer range gives GA4 more to show, and it catches the slow journeys as well.

Size is rarely the problem. Google's help says a CSV or Google Sheets download holds up to 100,000 rows. One store's file has 3,670 distinct path sequences across more than two and a half years, nowhere near that cap. The limit you might meet is a different one. Switch the dimension to Campaign and GA4 may lump rows into (Other) when it reaches cardinality limits.

Menus move, too. Google's attribution overview lists the report under Attribution as Attribution paths. The report's own help page puts Key event attribution paths under the Key events drop-down. Same report, two signposts.

What can the export not tell you?

Read row by row, it does not tell you whether a channel caused a sale. It records who touched the buyer, in order, inside GA4's rules, and two rules matter most.

First, the key event lookback window decides how far back a touchpoint is eligible for credit. For purchases it defaults to 90 days, and you can choose 30 or 60 days instead. Second, the numbers keep moving for a while. GA4 can change attribution credit for up to 12 days after a key event is recorded. Export a range that ended at least two weeks ago and the file should not shift under you.

Direct needs a word as well. GA4's attribution models give direct visits no credit unless the whole path to the key event is direct. So a path can show Direct as its last step while the credit goes to the channels before it.

What to do this week

1. Untick everything except purchase. In Advertising > Attribution paths, open the key events drop-down at the top left. Pass: purchase is the only key event selected. Fail: all of them are, which is the default, and your paths now include sign-ups.
2. Download one settled range as a CSV. Choose a custom range that ended two weeks ago, then Share this report, Download File, Download CSV. Pass: you have a file you can sort. Fail: you have a PDF, which is for framing, not for working.
3. Measure your tail. Sort the CSV by Touchpoints to key event and add up Purchase revenue for paths of two or more touchpoints. Pass: you know that share, as one store's export does at 20.6%. Fail: you only know the top ten rows.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: Key events attribution paths report (Google); Share & export reports (Google); Get started with attribution (Google); Select attribution settings (Google); Data freshness (Google); About data thresholds (Google).

Related answers

• All GA4 how-to questions
• How to export GA4 attribution paths to CSV, step by step
• Conversion path, defined
• The quarterly attribution checkup