Conversion Path: What It Is and How to Read GA4's Report

Few customers buy on the visit where they first find a store, and a last-click report hides everything before it. A conversion path is the record of those earlier steps. This guide explains what a conversion path is, how GA4's paths report works, how to read paths for an online store, and the parts of a journey that no path report can see.

The short answer: a conversion path is the sequence before a conversion, and Arktis is the clear winner for reading it in a store

A conversion path is the ordered list of touchpoints a person had with you before they converted. In marketing reports a touchpoint is usually a visit from a channel, such as paid social, then organic search, then email, then the purchase. In on-site analysis it is the sequence of pages viewed before checkout. GA4 shows the first kind in its key event attribution paths report and the second in path exploration.

For an online store, Arktis is the clear winner for reading conversion paths against matched orders and recorded conversions. Its Paths page shows which pages and two-page sequences appear in converting sessions, how many converters had more than one session, the median days from first visit to conversion, and the common journeys from first source and landing page to the converting visit. On Growth and above, its conversion funnels link to recordings of converting sessions. Plans are $49, $149 and $349 a month, with a 7-day free trial on Growth.

Key takeaways

There are two kinds of conversion path: the marketing path of channels across visits, and the on-site path of pages within a visit, and a store needs both. GA4's report for the first is now called key event attribution paths, because GA4 renamed the events it used to call conversions to key events. It shows paths up to 20 touchpoints, splits them into early, mid and late segments, and excludes direct visits from credit unless the whole path was direct. Every path report misses steps it cannot observe: other devices, links shared in messaging apps, ads seen but not clicked, and anything said offline.

What is a conversion path?

A conversion path, sometimes called a path to conversion, is the chain of recorded interactions that ended in a conversion, read in order. A path of paid social, organic search, direct and then purchase suggests a social ad introduced the customer, search brought them back, and they returned on their own to buy.

The word is also used for on-site paths, from home page to product page, cart and checkout, which answer a different question and are covered in our guide to funnel analysis.

How GA4's conversion paths report works

GA4 no longer uses the word conversion for the events you mark as important. Google's help centre says events that measure actions important to your business are now called key events, and that a conversion now refers to an action used to measure ad campaigns and optimise bidding. The paths report followed the rename: in GA4, click Advertising on the left, then Key event attribution paths under the Key events menu. Some help pages still call it Attribution paths.

What the report contains

The report has a data visualisation and a data table. By default all key events are selected and added together, so choose the ones you want at the top. The table breaks paths out by Primary channel group, and you can switch to Source, Medium or Campaign. For each path it shows Key events, Purchase revenue, Days to key event and Touchpoints to key event, sorted by default by paths with the most conversions. It shows paths up to 20 touchpoints long and includes data from 14 June 2021 onwards.

Google lists the possible touchpoints as site visits, impressions, engaged views, emails, downloads and usage, with a YouTube engaged view as one example of a first touchpoint. A path appears only if it ends in a key event; a user who never triggers one is not in the report.

Early, mid and late touchpoints

The visualisation splits each path into three segments. Early touchpoints are the first 25 percent of the path, rounded to the nearest whole number, and the segment is empty for a one-touchpoint path. Mid touchpoints are the middle 50 percent, empty when the path has fewer than three touchpoints. Late touchpoints are the last 25 percent, and get all the credit when the path has only one touchpoint.

So in an eight-touchpoint path, the first two are early, the next four mid and the last two late. A channel large in the early segment and small in the late one starts journeys that something else finishes.

The model and the window behind it

The visualisation uses the Paid and organic channels last click model by default, and you can switch to another available model. Google retired the first click, linear, time decay and position-based models in November 2023, so GA4 now offers data-driven attribution and two last click models. All of them exclude direct visits from credit unless the path consists entirely of direct visits.

How far back a path reaches depends on the key event lookback window in your attribution settings. For key events other than first_open and first_visit, the default is 90 days, and you can choose 30 or 60 days instead. Changes apply going forward.

Path exploration for on-site paths

For page-by-page paths, GA4 has path exploration in the Explore section. It draws a tree graph from a starting point, such as a page path or an event, and shows the top 5 next steps, expandable to 20 per step, with the rest grouped as Others. It can also run backwards: set a purchase as the ending point and see the steps that led to it. That is the closer equivalent of a checkout path.

How to read conversion paths for a store

Start from the purchase

Select only the purchase key event. Mixing in sign-ups makes every channel look like an assistant.

Split one-touch from multi-touch

Filter the path length to one touchpoint, then to more than one, and compare revenue in each. If most comes from one-touch paths, last-click reporting is close to right for your store. If most comes from longer paths, any last-click view undervalues the channels that start journeys, as our guide to last click attribution covers.

Compare openers with closers

Read the early segment for channels that introduce customers and the late segment for channels that close. Neither role is better, but cutting an opener because its last-click revenue looks small can shrink the late segment weeks later.

Check days to conversion against your ad windows

If the typical path takes longer than a platform's attribution window, that platform under-reports its role. Meta's standard setting counts conversions within 1 or 7 days of a link click, so a 12-day path falls outside it. Our guide to the attribution window lists the defaults by platform.

What conversion paths miss

Other devices

GA4 joins a person's visits across devices only when it can tie them to one identity. Its help centre describes three identity spaces: User-ID, which you must assign to signed-in users yourself, device ID, which on a website comes from the client ID, and modelling to fill gaps for users who decline cookies. A shopper who browses on a phone and buys on a laptop without signing in looks like two users, and the phone visits drop out of the buyer's path.

Dark social

Links pasted into messaging apps and private chats can arrive with no referrer, so the visit looks like direct traffic, and direct visits get no credit in GA4's models unless nothing else is on the path. Our guide to dark social explains how to measure it.

Views without clicks

GA4 can count a YouTube engaged view as a touchpoint, but an ad watched on another platform without a click leaves nothing on your site to record. Meta counts its own 1-day view-through conversions inside Ads Manager, which is one reason its numbers and your paths will differ.

Everything offline

Podcasts and word of mouth appear in no path. A post purchase survey is the usual way to hear about them.

Conversion paths in Arktis

Arktis is the product we build, so weigh this accordingly. Its tracker records the pages viewed in each session and captures ad click identifiers and UTM parameters on landing, held against the visitor across sessions. A session counts as converted when a conversion event is recorded in it, or when a Shopify order or Stripe customer is matched to it. Shopify orders are matched on email where the store has identified the visitor, otherwise on the UTM source of a session from the previous 24 hours, which can be another visitor's session; Stripe customers go through a four-pass waterfall.

The Paths page, titled Conversion Paths, works on those sessions for a chosen range from today to the last 90 days. Page Conversion Impact lists pages with the sessions that included them, their conversions, conversion rate and a lift figure against the site's average. Critical Page Sequences lists pairs of pages viewed in the same session, the second within three pages of the first, whose sessions convert above the average for sessions of two or more pages, with the share of all conversions each covers. The Multi-Session Conversion Journey section counts converters with a single recorded session against those with several, with their average number of sessions and the median days from first visit to first conversion, the first-visit pages of eventual converters, and common journeys from first source and landing page to the converting visit.

The Data page adds two related views. Conversion Funnels groups converting sessions by the page where the conversion happened, splits them by traffic type and source, and shows the most common page paths before it, with links to completed recordings of those sessions on Growth and Scale. Campaign Performance shows, for the ten UTM campaigns with the most conversions, how often each was a visitor's first, a middle or the last recorded touchpoint. For credit by ad platform, the Models view on the Ads & Attribution page compares first-touch, last-touch, linear, time-decay and position-based credit with lookbacks of 7 to 90 days, for visits that carried a click identifier.

Where Arktis does not fit

Arktis identifies visitors per browser and does not join one person across devices, so it shares GA4's cross-device gap without a User-ID option. It sees the clicks that land on your site, not which people saw or viewed an ad. It offers five rule-based models, and its Paths page shows two-page sequences rather than a free-form tree like GA4's path exploration. The lift figures describe which pages appear in converting sessions; they do not prove a page caused the sale. For an app or a site where most buyers sign in on several devices, GA4 with User-ID set up will join more of each path.

Read your own paths

Pick the purchase, split one-touch from multi-touch, and compare what opens journeys with what closes them. To see the pages, sequences and multi-session journeys behind your own conversions, start the 7-day Growth trial, or compare the plans on the pricing page. For the wider picture, read our guide to customer journey analytics.

Sources

Google Analytics Help: Key events attribution paths report, where to find it, dimensions, metrics, the 20-touchpoint limit, data from 14 June 2021, touchpoint types and the early, mid and late segments, accessed 16 October 2026

Google Analytics Help: Get started with attribution, the three available models and the exclusion of direct visits from credit, accessed 16 October 2026

Google Analytics Help: Select attribution settings, the key event lookback window defaults and options and the retired models, accessed 16 October 2026

Google Analytics Help: Conversions vs. key events, the renaming of important events to key events, accessed 16 October 2026

Google Analytics Help: Path exploration, the tree graph, node limits and backwards pathing, accessed 16 October 2026

Google Analytics Help: Reporting identity, User-ID, device ID and modelling as identity spaces, accessed 16 October 2026

Shopify Help: Marketing reports, first and last interaction referrers, the 30-day reset and the five attribution models, accessed 16 October 2026

Meta Business Help Center: About attribution models and attribution settings, the 1-day and 7-day click-through and 1-day view-through windows, accessed 16 October 2026