
The short answer
GA4 shows "(not set)" as a landing page when a session has no page_view event. That's Google's own explanation, word for word on its (not set) help page. The Landing page dimension is "the page path associated with the first page view in a session," so a session that starts with a click, a form submit or a background engagement hit, and never gets a page view, has nothing to report.
The useful question is which visitor paths produce those sessions. Most guides list five or ten possible causes and leave you to guess. We ran nine paths through the real Google tag and watched what it sent. Two causes showed up clearly, one popular suspect turned out to drop data rather than create (not set), and the biggest cause is something you can shrink but never fully remove.
How we tested it
We built a two-page fixture site, loaded the real gtag.js from googletagmanager.com for a made-up measurement ID, and drove it with Playwright in headless Chromium 148. Every request to /g/collect was intercepted on our machine and answered locally, so no hit reached Google or anyone's property. To simulate idle time without waiting half an hour, we installed Playwright's fake clock and fast-forwarded it 31 minutes.
For each hit we logged the event name (en), the session ID (sid), the session count (sct) and the _ss flag, which marks the event that starts a session. That flag is the whole story here: whichever event carries _ss=1 opens the session, and if that event isn't a page_view and no page_view follows, the landing page is (not set).
One limit up front: we can see exactly what the tag sends, not how GA4 processes it on Google's side. Where the processing matters, we lean on Google's documentation and say so.
What the Google tag sent in 9 visitor paths
- Normal visit:
page_viewcarried_ss=1. Landing page set. This is the baseline. - Idle 31 minutes, then a click on the same page: a new session (
sctwent from 1 to 2) opened withgenerate_leadcarrying_ss=1. No page_view in that session. (not set), and the lead sits inside it. - Switched to another tab for 31 minutes, came back, clicked: identical to the line above. New session, opened by the click, no page_view.
- Idle 31 minutes, then went to a second page: the new session opened with a
user_engagementhit fired from the old page as the visitor left, then the second page'spage_viewarrived in the same session. That session does get a landing page, but it's page two, not where the visitor actually was. send_page_view: falsein the config, then a lead click: the session opened withgenerate_leadand no page_view ever followed. On a site set up this way, every session is (not set).- An event pushed before the config command: it was never sent at all. Only the page_view went out. Not a (not set) cause, but a silent data loss worth knowing about.
- Consent banner, accepted with
consent update: the page_view went out cookieless (gcs=G100), and within milliseconds of the accept click the tag sent auser_engagementhit with consent granted (gcs=G111). - Consent banner, accepted with a second
consent default: same cookieless page_view, but no user_engagement on accept. The first granted hit was the lead event, seconds later. - Back button after 31 minutes: in headless Chromium the page reloaded and sent a fresh page_view, so no (not set). A real browser restoring the page from its back/forward cache may behave differently; we didn't get a cache restore in this run, so we're not making a claim either way.
The takeaway: the sessions that end up as (not set) are the ones where something other than a page view gets to _ss=1 first and no page view follows. Everything below is about stopping that from happening.
Cause 1: idle tabs crossing the 30-minute timeout
This is the one nobody can fully fix. By default a GA4 session times out after 30 minutes of inactivity. People leave tabs open all the time: a quote page during lunch, a pricing page while they ask their partner. When they come back and click without reloading, the tag correctly starts a new session, but there's no page load to send a page_view.
What makes this painful is that it's the returning, interested visitor who does it. In our test the lead event itself opened the new session. In your reports that conversion is credited to a session with landing page (not set), so the page that actually earned the lead gets no credit for it. If you've been chasing a mismatch between GA4 and your ad platforms, this is one of the quiet contributors; we covered the bigger ones in why Google Ads and GA4 conversions don't match.
Fix: raise the session timeout. In GA4 go to Admin, Data streams, pick the web stream, Configure tag settings, Show more, Adjust session timeout. A longer timeout means fewer returning visitors cross it. The trade-off is real: some visits that used to count as two sessions become one, so session counts dip and engagement rate shifts. Annotate the date.
We wouldn't try to "fix" this with a script that fires a page_view whenever a tab becomes visible again. That inflates page views and makes every other page report less trustworthy to repair one dimension.
Cause 2: page_view switched off, or firing late
send_page_view: false shows up in a lot of setups for good reasons: single-page apps that send page views on route changes, sites that wait for a consent decision, or a developer who wanted to add custom parameters first. The problem is when the manual page_view never comes, or comes after something else.
In our test, with page_view off and a lead button on the page, the session opened with the lead event and had no landing page at all. The same thing happens in Google Tag Manager when an event tag (a scroll tag, a form-start tag, a chat-widget tag) fires on an earlier trigger than the Google tag's page view.
Fix:
- gtag.js: if you set
send_page_view: false, sendgtag('event', 'page_view')yourself right after the config, before any other event can fire. - Google Tag Manager: fire the Google tag on Initialization or the page-view trigger, and don't give event tags an earlier trigger like Consent Initialization.
- Single-page apps: confirm a page_view goes out on the very first load, not only on route changes. This is the most common gap we see on React and Next.js sites.
The related trap from our test: an event pushed to the dataLayer before the config command was dropped entirely. It wasn't queued, it just never went out. If a script on your page fires gtag('event', ...) above the config line, move it below.
Cause 3: consent banners that call "default" twice
Google's help page names this one directly: using the default command to record the visitor's choice after they click Accept, instead of update, means "the user_engagement event isn't sent, and the initial event with the session_start parameter is lost." Our wire capture matches the first half of that sentence. With update, the tag fired a granted user_engagement hit immediately on accept. With a second default, it fired nothing until the next real event.
What GA4 does with those hits on its servers we can't see from the browser, but the documentation is clear about the result, and the fix is one word. If your banner is a plugin, check its Google consent mode integration setting rather than its marketing page. We walked through the same gcs parameter checks in our consent mode troubleshooting guide.
Fix: gtag('consent', 'default', ...) once, before the config, with your defaults. gtag('consent', 'update', ...) when the visitor decides. Never a second default.
Cause 4: server-side events that arrive on their own
This one isn't in our browser test, so treat it as a known pattern, not a measurement. If your CRM, checkout or an n8n workflow sends events to GA4 through the Measurement Protocol, those events don't carry a session ID unless you add one. Google's sessions page says the identifiers are "not included automatically in events from Measurement Protocol." An offline conversion with no session to join, or with a session ID that has already timed out, can end up in a session that never had a page view.
Fix: capture the ga_session_id and client ID on the site when the lead is created, store them with the lead, and send both with the server event. If the event happens days later (a closed deal, a signed contract), accept that it's an offline conversion and report it separately instead of forcing it into web sessions.
How to find out which cause you have
Don't fix all four blind. Spend 15 minutes on diagnosis first.
- Measure the share. Open Reports, Engagement, Landing page. Note (not set) sessions divided by total sessions for the last 28 days.
- See what's inside those sessions. Create a free-form Exploration filtered to Landing page exactly matches (not set). Put Event name in rows and Event count as the metric. A mix of
user_engagement,scroll, clicks and the odd lead points at idle tabs. The same form or custom event showing up in nearly every one of those sessions points at a tag order problem (Cause 2). - Break it down by page. Add Page location to the same Exploration to see which pages those events happened on. If one template or one app route dominates, you've found your broken page.
- Open DevTools on that page. Filter Network to
collect, reload, and check that the first hit'senispage_view. If it's anything else, that's Cause 2.
If you have the BigQuery export, one query settles it:
SELECT
first_event,
COUNT(*) AS sessions
FROM (
SELECT
user_pseudo_id,
(SELECT value.int_value FROM UNNEST(event_params) WHERE key = 'ga_session_id') AS sid,
ARRAY_AGG(IF(event_name IN ('session_start', 'first_visit'), NULL, event_name)
IGNORE NULLS ORDER BY event_timestamp LIMIT 1)[SAFE_OFFSET(0)] AS first_event,
COUNTIF(event_name = 'page_view') AS page_views
FROM `your_project.analytics_XXXXXX.events_*`
WHERE _TABLE_SUFFIX BETWEEN '20261001' AND '20261031'
GROUP BY user_pseudo_id, sid
)
WHERE page_views = 0
GROUP BY first_event
ORDER BY sessions DESC;
It lists sessions with zero page views grouped by the event that opened them (skipping the session_start and first_visit rows the export adds). That's the same _ss story from our test, read from your own data.
Our rule of thumb: if (not set) is a few percent of sessions and the opening events are mostly engagement and scroll hits, you're looking at idle tabs. Raise the timeout and move on. If it's into double digits, or it's concentrated on your conversion events, something in the tag setup is wrong and it's worth an hour of a developer's time.
What fixing it won't do
GA4 doesn't reprocess sessions it has already processed, so last month's (not set) rows stay in last month's reports. Annotate the day you ship the fix and compare the share before and after. If you're cleaning up a GA4 property for a client report, our technical SEO audit checklist covers the other measurement checks we run alongside this one, and what a tracking audit includes explains how we scope them.
Seeing (not set) on the landing pages that drive your leads? We'll check your Google tag, GTM order, consent setup and server-side events, find which cause is creating sessions without a page view, and fix it so your conversions are credited to real pages again. Get a Free Automation Audit and we'll email you the initial audit report within 24 hours.
