What does the conversion paths report in GA4 tell me?

It usually tells you three things. First, the order in which channels reached buyers before a purchase. Second, how many touches and days that took. Third, how one attribution model splits credit between early, middle and late touches. It does not tell you which touch caused the sale.

If you went looking for a report called conversion paths, GA4 now files it under another name. Google calls important events key events these days, not conversions. The paths report sits under Advertising, as Attribution paths. Google Ads, meanwhile, keeps a Conversion paths tab of its own, under Attribution in the Goals menu, built from ad interactions. This answer is about the GA4 one.

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 these rows next to the chart at the top of the GA4 report. Google splits every path into early, mid and late touchpoints. Early is the first 25% of a path's touches, mid the middle 50% and late the last 25%. With one touch, the early segment is empty and the late segment takes all the credit.

In one store's export, journeys with 1 touch hold 79.5% of revenue on the Journeys sheet. In a chart built on Google's rules, all of that revenue lands in late, under any model. The mid segment stays empty below three touches, so it only ever sees the longer rows.

So the chart's story about which channels open and assist journeys rests on the other three rows. On the Journeys sheet, they hold 20.6% of revenue together (12.2% + 5.4% + 3.0%). That is the slice where the choice of model can change anything.

Those rows are also where the waiting happens. On the Journeys sheet, journeys of 2 to 3 touches took 12.5 days to buy, against 0.5 days for a single touch. Journeys of 10 or more touches on the Journeys sheet took 16.0 days, a shade quicker than the 16.9 days of 4 to 9 touches. More touches did not mean more days: the busiest journeys packed their touches closer together.

What the export cannot show is cause. It lists journeys by length and their shares of revenue. Nothing in it says whether a buyer would have bought without one of the touches.

Why does the usual reading mislead?

The usual reading treats the report as the customer journey, start to finish. Three details get in the way.

Only buyers appear. In Google's words, without a key event there is no path to show. Its own example leaves out a five-touch journey that never ended in one. So the report shows how buyers travelled, never how often a channel's visitors left with nothing. A channel can look like a great opener simply because it opens a lot of journeys.

The default chart barely looks back. It starts on the Paid and organic last click model. That model ignores direct traffic and gives the whole sale to the last channel clicked. Under it, an early or mid touch only earns credit when direct visits close the path.

The default key events are a mix. GA4 selects every key event in the report and adds them together. With ads personalization on and a Google Ads account linked, GA4 also marks add_to_cart, begin_checkout, session_start and view_item as key events. Leave those ticked, and your paths include people who only looked around.

What can the report not tell you?

It cannot tell you whether a touch caused anything. A path is a list of who showed up, in order. Showing up is not the same as making the sale happen.

It cannot fully tell you what the data-driven model saw. Switch the chart to data-driven and credit spreads across more touches. Google says that model weighs converting and non-converting paths, using a counterfactual approach. It still sees only the touchpoints GA4 recorded, and the table shows each row's credit as an average across its paths.

And it cannot see anything GA4 never recorded. A podcast mention, a shop window or a friend's tip leaves no touchpoint. The buyer arrives later by typing your address, and the path starts with Direct.

To learn what a channel caused, switch it off for part of your audience and compare sales with the rest. That is a holdout test, and the paths report cannot run one for you.

What to do this week

1. Measure the revenue no model can split. In GA4, click Advertising, then Attribution paths, and set the path length filter to equal to 1 touchpoint. Compare Purchase revenue in the top row with the unfiltered total. Pass: you know your one-touch share, as one store's export does at 79.5%. Fail: you argue about models over revenue that no model can move.
2. Name your openers. Use the attribution model drop-down above the chart to switch from Paid and organic last click to Data-driven, with purchase as the only key event. Pass: you can list the channels whose early-touchpoint credit grows. Fail: nothing moves, which can mean most of your revenue sits in one-touch paths.
3. Check that every paid channel shows up. Switch the dimension drop-down above the table to Source, then to Medium, and look for each platform you pay. Pass: each one appears somewhere in your paths. Fail: a paid channel never appears, so check the UTM tags on its links before you judge it.

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 Analytics Help); Get started with attribution (Google Analytics Help); Mark events as key events (Google Analytics Help); Conversions vs. key events in Google Analytics (Google Analytics Help); About attribution reports (Google Ads Help).

Related answers

• All GA4 how-to questions
• How to read the GA4 conversion paths report, step by step
• What do GA4 conversion paths tell a sports nutrition store?
• How do I export attribution paths from GA4?
• Causal attribution, explained