New vs Returning Visitors: GA4 Definitions, Ratios and Real Value

New vs returning visitors splits your traffic into browsers your analytics tool has never seen before (new) and browsers it recognizes from an earlier visit (returning). In Google Analytics 4 the split is made by the _ga cookie, so it counts browsers on devices, not people. The ratio on its own tells you little; it becomes useful when you compare what each group converts into, by channel and landing page.

This guide covers how GA4 decides who is new, why the numbers never add up, where to find the report, what a “good” ratio really means, a worked example that turns the split into dollars, and a checklist for when your new visitor share looks wrong.

What New and Returning Visitors Mean

• New visitor: a visit from a browser that carries no identifier from your site. The tool creates one and files the visit under “new.”
• Returning visitor: a visit from a browser that already carries your identifier from an earlier session, so the tool links it to that history.

Three consequences follow from that definition, and most misreadings of the report come from forgetting one of them:

1. A visitor is a browser, not a person. The same person on a phone and a laptop is two visitors. The same person in Chrome and then Safari on one laptop is also two.
2. “New” means “not recognized.” Someone who cleared cookies, declined your cookie banner or browsed in a private window is new again, however many times they have visited.
3. Returning is not the same as loyal. A customer checking an order status five times is returning. So is your own team. A returning visitor is not a returning customer; repeat purchase rate is a separate metric that lives in your store or CRM.

How GA4 Decides Who Is New and Who Is Returning

GA4’s tag sets a first-party cookie called _ga that holds a random client ID. When a page loads without that cookie, GA4 creates one and logs a first_visit event. When the cookie is already there, the visit belongs to an existing user.

Google’s user metrics documentation defines the four numbers you will see:

The _ga cookie has a default expiry of 2 years, and every visit refreshes it. If your site has users who log in, sending a User-ID lets GA4 connect their devices, which turns some of those duplicate “new” visitors back into one returning user. Anonymous visitors stay browser-based.

A common claim that GA4’s definitions contradict

Several guides say that a person who visited six months ago shows up as a new visitor in a 30-day report because their earlier visit falls outside the date range. That is not how GA4 works. New users are users whose first_visit event happened in the range. If the person’s cookie survived, their first visit happened six months ago, so in this month’s report they are a returning user. They only appear as new if the cookie was lost in between, which is far more common on some browsers than others (see the Safari section below).

Why New + Returning Users Don’t Add Up to Total Users

This is the most common complaint about the report, and it is expected behavior. The split is made per session, while the user counts are de-duplicated across the whole range. A user who first visits on the 3rd and comes back on the 20th has one new session and one returning session in the same month, so they are counted in both rows but only once in Total users.

The gap grows with longer date ranges and with sites people revisit quickly. Shorten the range to one day and the two rows add up almost exactly. Other reasons the numbers drift apart:

• Inactive users. A returning visitor who leaves within seconds without engaging counts in Total users but not in Active users (“Users”), so comparing new users against the Users column looks wrong even when it isn’t.
• Thresholding. When Google signals is on, GA4 may hide small rows to protect privacy, which removes users from some rows and not others.
• Different scopes in one table. Mixing user-scoped and session-scoped dimensions in an exploration produces totals that do not match the standard reports.

If you need one person, one label for a period, export GA4 to BigQuery and define it yourself, for example “new this month if the first session date is in this month.” For most marketing decisions the standard split is enough once you know why it overlaps.

Where to Find New vs Returning Visitors in GA4

Universal Analytics had a dedicated report under Audience, Behavior, New vs Returning. That report is gone. In GA4 you have three routes:

1. Retention overview. Reports, Retention. The first card charts new users and returning users over time. Good for trends, not for analysis.
2. A free form exploration. Explore, Free form. Add the dimension New / returning to Rows. Add Total users, Sessions, Engagement rate, Key events and Total revenue (or Event value) to Values. Then add a second dimension such as Session default channel group, First user default channel group or Landing page + query string.
3. The New / established dimension. Also in Explorations. Google’s dimensions and metrics reference defines “new” here as users who first visited within the last 7 days. It answers a different question than New / returning, so do not mix the two in one report.

Visitors, Unique Visitors and Users

Older tools and hosting dashboards use different words for the same ideas:

Unique visitor counts are tied to their period. Thirty daily counts added together are larger than the monthly count, because anyone who visited on two days is unique on each day but only once in the month. Server log tools that estimate uniques from IP address and browser string will also disagree with GA4, since they count bots and shared IPs differently.

How New and Returning Visitors Behave

On most sites the two groups behave differently, which is the main reason to split them before reading any engagement or conversion report:

This also means a blended average hides both groups. If your engagement rate fell, check whether returning visitors engaged less or whether the mix simply shifted toward new visitors. Our guide to the GA4 engagement rate walks through the same mix effect.

What Is a Good New vs Returning Visitor Ratio?

There is no universal good ratio, and the published benchmarks prove it. Among the pages ranking for this topic, one says a healthy ecommerce site has around 30% returning visitors, another says ecommerce sites run 55% to 75% new, and a third quotes an agency seeing a 25/75 split. None of them cites a primary dataset you can check, so treat any single number as an anecdote.

What actually sets your ratio:

• Business model. Publishers, SaaS apps and stores with repeat purchases depend on people coming back. Lead generation sites and one-off purchases depend on a steady flow of new visitors.
• Channel mix. A month with a strong SEO launch or a paid campaign pushes the new share up. A month with heavy email pushes returning up. Neither is a change in loyalty. Split paid from organic when you read it, since paid spend behind a page inflates its new visitors in a way organic growth does not.
• Logged-in and utility traffic. Customer logins, order tracking, documentation and support pages produce returning visits that have nothing to do with marketing.
• Browser mix and consent rate. The more of your traffic loses its cookie, the more “new” you look (next section).

The useful benchmark is your own trend, read next to conversions. A ratio that moves while conversions per visitor stay flat is a traffic mix change. A ratio that moves together with value is a signal.

The Safari 7-Day Cap: Why Your New Visitor Share Depends on Browser Mix

None of the usual explanations mention the biggest technical driver. Google’s cookie usage page for GA4 states that browsers limit how long first-party cookies live when the user does not return: a maximum of 400 days in Chrome and 7 days in Safari. GA4 asks for 2 years, but on Safari a visitor who comes back after more than a week is usually a new user again.

So a site with a large iPhone and Mac audience reports more new visitors and fewer returning visitors than a similar site used mostly on Chrome, even with identical loyalty. B2B sites with monthly buying cycles are hit hardest, because buyers often come back after more than seven days. Our guide to Safari ITP explains the browser rules behind this.

How to check it on your site: in a free form exploration, put Browser in Rows and New / returning in Columns, with Sessions as the value. If Safari’s new share is much higher than Chrome’s, part of your “new” traffic is returning visitors whose cookie expired. Compare trends within each browser, not the blended ratio.

Worked Example: Turning New vs Returning Into Conversions and Value

The ratio becomes useful when each group carries a conversion count and a value. All numbers below are illustrative, for a B2B service site over one month, with each lead worth $150 (see how to calculate conversion value to set your own).

How to read it, step by step:

1. The overlap. 7,200 + 3,400 = 10,600, which is 600 more than the 10,000 total users. Those 600 users had their first visit this month and came back later in the month.
2. The rates. 72 / 7,200 = 1.0% for new and 102 / 3,400 = 3.0% for returning. Returning visitors convert three times as often here.
3. The value. 72 × $150 = $10,800 and 102 × $150 = $15,300. Per user, that is $1.50 for a new user and $4.50 for a returning one.

The tempting conclusion is that returning visitors are the valuable ones and new visitor channels underperform. Now add the dimension First user default channel group to the returning row, which shows the channel that brought each user the first time, and compare it with the channel of the session in which they converted:

In this example, the returning visitors are mostly people your SEO pages brought in as new visitors weeks earlier. Cutting the content that attracts new visitors because “new visitors don’t convert” would shrink the returning group a month later. The same pattern shows up with AI assistants: someone clicks through from ChatGPT, leaves, and returns directly to buy. Read first touch attribution for the full method.

Finally, split the new row by Landing page + query string. Pages that bring many new visitors with near-zero leads, even after they return, are traffic without a business result. Pages whose new visitors convert later are working, even if their same-session conversion rate looks poor. Doing this by landing page and channel is the core of SEO conversion tracking.

Reading Changes in the Ratio: A Decision Table

When the New Visitor Share Looks Wrong: Debug Checklist

An inflated new user share is usually a recognition problem, not a marketing one. Work through these in order:

Note that tools built without cookies handle this differently. Some cookieless analytics tools rotate their anonymous identifier every day, so they cannot say whether someone came back next week at all. Check how a tool identifies visitors before you compare its returning visitor numbers with GA4.

How to Bring Visitors Back (and Convert Them the First Time)

Optimizing for returning visitors starts on the first visit. The tactics that work fall into two groups:

For new visitors

• Answer the query the page ranks for. A visitor who got what they searched for is more likely to come back by name.
• Give a next step that matches their stage. A reader of a how-to guide wants a related guide or a newsletter, not a demo form. A visitor on a comparison or pricing page may be ready for the demo.
• Show who you are quickly. Clear navigation, a plain explanation of what you sell and visible proof such as reviews or certifications.
• Capture a way to reach them. Email signup, a useful download or an account. Without it, the return visit depends on them remembering you.

For returning visitors

• Email. Newsletters, content updates and cart or quote reminders. Tag every link with UTM parameters so the return visit is not filed as Direct.
• Retargeting. Show people who already visited a message that fits what they saw, not the same intro ad new audiences get.
• Social and push notifications. Useful for publishers that put out new content often.
• Make the return trip fast. Returning visitors head for pricing, login, documentation or the product they viewed. Keep those one click away, and personalize when you can (recently viewed items, saved carts).

Branded search is a good proxy for how many people come back by choice. Our guide to branded vs non-branded traffic shows how to split it.

Measure What Each Group Is Worth

The new vs returning split answers “who came.” Decisions need “what did they do, and what was it worth.” Put a value on each conversion type, then read value per visitor by channel and landing page, with the first visit channel next to the converting one. SEOConversion reports conversions and their value by landing page for organic search and AI assistants, which covers the second half of the worked example above for those two channels.

FAQ