5 GA4 Explorations Every SEO Should Build

GA4’s standard reports tell you how much organic traffic you got. They rarely tell you which pages earn it, where those visitors go next, or whether they come back. For that kind of GA4 explorations SEO work, you need the Explore section. In my experience, five explorations cover most of what an SEO needs to answer each month.

In this guide I give you the exact build steps for each one: an organic landing page report, an organic path exploration, a comparison of organic and AI assistant visitors, a funnel from organic entry to a key event, and a cohort view of organic users. At the end I cover the limits that catch people out: retention, sampling and thresholds.

Before You Build: Three Things to Set Up

Explorations are only as good as the data under them. I always check these three settings first. Each one quietly changes what your explorations can show.

• Data retention. Explorations only use event data inside your retention window (how long GA4 keeps detailed data). New properties keep just 2 months. Go to Admin > Data collection and modification > Data retention and set event data to 14 months, the maximum for standard properties. It doesn’t apply to past data, so do it now.
• Key events. Several explorations below measure results. Mark your form submissions, sign-ups, calls or purchases as key events first. My guide to setting up GA4 key events covers which ones to pick.
• A saved organic segment. You’ll reuse the same segment in four of the five builds. Create it once (steps below) and it will be there in every new exploration.

How to Create the Organic Sessions Segment

1. Open Explore and start a Blank exploration.
2. In the Variables column, click the + next to Segments and choose Session segment.
3. Name it “Organic Search sessions”.
4. Add a condition: Session default channel group exactly matches Organic Search.
5. Click Save and apply.

A session segment keeps only the sessions that came from organic search. A user segment would keep every session from anyone who ever came in through organic search. That makes the numbers look bigger than they are. For most SEO questions, I stick with session scope. For how Google now sends result clicks, see my note on the google.com/goto redirect.

Keep in mind that clicks from Google’s AI Overviews and AI Mode come from google.com. So GA4 counts them as Organic Search, and this segment includes them.

1. Organic Landing Pages With Engagement and Key Events (Free Form)

This is the exploration you’ll open most often. It’s the one I open first every month. It shows which pages bring in organic visitors, and whether those visitors do anything useful once they arrive.

Build Steps

1. Go to Explore > Free form. Rename the exploration “SEO: organic landing pages”.
2. Set the date range to the last 3 months, or the last 28 days if you have a lot of traffic.
3. Under Segments, import your Organic Search sessions segment. Drag it into the Segment comparisons box in the Settings column.
4. Under Dimensions, click + and import Landing page + query string and Device category.
5. Under Metrics, import Sessions, Engaged sessions, Engagement rate, Average engagement time per session, Key events and Session key event rate. Add Total revenue if you run an online shop.
6. Drag Landing page + query string into Rows. Set Show rows to 100 or more.
7. Drag the metrics into Values. Then sort by Sessions, highest first.

Drag Device category into Columns when you want to see if mobile visitors act differently on the same page. Keep the table to one or two row dimensions. Every extra dimension splits the data into smaller rows that are harder to read. If you’re not sure what each metric counts, my guide to GA4 engagement metrics explains them.

How to Read It

• High sessions, low engagement rate: the page ranks but doesn’t satisfy. Check if it matches what searchers want, and if the answer sits near the top.
• Healthy engagement, zero key events: the content works, but there’s no clear next step. Add an internal link, a product link or a contact option.
• Low sessions, high key event rate: your hidden winners. These are the pages I’d prioritise. They deserve better internal links and a content refresh to grow their rankings.

Watch for a row called (not set). It usually means a session started without a page view. For example, a session times out and restarts on an event. A small share is normal. A large share points to a tagging problem you should look into.

If you want this as a regular view your team can open without touching GA4, my Looker Studio SEO dashboard guide shows how to rebuild it next to Search Console data.

2. Organic Path Exploration

A path exploration shows the order of pages or events after a starting point. For SEO, it answers a simple question: when someone lands on a page from search, where do they go next?

Build Steps

1. Go to Explore > Path exploration. GA4 loads a sample path, so click Start over in the top right.
2. Import the Organic Search sessions segment and add it to Segment comparisons.
3. Click Starting point and choose Page path and screen class. Pick one of your top organic landing pages from exploration 1.
4. In the Step +1 column, set the node type to Page path and screen class. Now the next steps show pages, not events.
5. Turn on View unique nodes only in the Settings column. This hides repeated page views caused by refreshes or tag quirks.
6. Click any node in Step +1 to open the step after it. Keep going for three or four steps.

You can also work backwards. Click Start over, choose Ending point instead, and pick your thank-you page or a generate_lead event. GA4 then shows the pages organic visitors passed through before they converted. I find this backwards view is often more useful than the forward one.

What to Look For

• Dead ends. If most sessions stop after the landing page, the page has no clear next step. That’s a sign to review your internal linking and topic clusters.
• Unexpected detours. Many visitors jumping from a blog post to your pricing page suggests a buying angle the post could serve better.
• Loops back to search. Visitors who go straight to your site search page often didn’t find what the landing page promised.

Path explorations get messy fast. My rule is one starting page at a time. Choose a date range long enough to give you a few hundred sessions.

3. Segment Comparison: Organic Search vs AI Assistant Referrals

Since 13 May 2026, GA4 has a built-in AI Assistant channel in the default channel group. Google’s What’s new in Google Analytics page announced it as a way to see visits from chatbots such as ChatGPT, Gemini and Claude.

When a referrer (the site a visitor came from) matches a known AI assistant, GA4 sets the medium to ai-assistant and the campaign to (ai-assistant). It then groups the session under AI Assistant. There’s nothing to set up. Before this, those visits landed in Referral. Or they landed in Unassigned when ChatGPT’s utm_source=chatgpt.com tag arrived without a medium.

Google hasn’t published the full list of assistants it recognises. So I check Session source inside the AI Assistant channel now and then, to see which ones it really catches on the site.

Build Steps

1. Create a second session segment called “AI Assistant sessions” with the condition Session default channel group exactly matches AI Assistant.
2. Start a new Free form exploration called “SEO: organic vs AI”.
3. Import both segments and drag both into Segment comparisons. Each segment gets its own set of columns.
4. Import Landing page + query string and Session source as dimensions.
5. Import Sessions, Engagement rate, Average engagement time per session, Key events and Session key event rate.
6. Put Landing page + query string in Rows and the metrics in Values.
7. Set the date range to start on or after 13 May 2026.

Covering Dates Before 13 May

Google hasn’t said if the new channel applies to older data. If you need a longer trend, build the AI segment on Session source matches regex instead. Regex is a text pattern that matches many values at once. For example:

.*(chatgpt|openai|perplexity|gemini.google|copilot.microsoft|claude.ai).*

If you already built a custom AI channel, as I cover in my guide to tracking ChatGPT, Perplexity and Gemini visits in GA4, you can use that channel in the segment condition instead. Custom channel groups work on past data.

How to Read the Comparison

Compare rates, not volumes. AI assistants will almost always send far fewer sessions than Google search. So the useful question is whether those sessions engage and convert at a similar rate.

Look closely at pages that show up in the AI columns but rank poorly in organic. They often have a clear definition, a comparison or original data that assistants like to cite. Swap Landing page for Session source in Rows to see which assistant sends which kind of visitor.

For a quick extra view, use the Segment overlap technique with the same two segments, switched to user scope. It shows how many people came in through both channels during the period.

4. Funnel From Organic Entry to a Key Event

A funnel exploration shows how many organic visitors finish each step towards a key event, and where they drop out. In my view, it’s the clearest way to show that SEO traffic leads somewhere.

Build Steps

1. Go to Explore > Funnel exploration and name it “SEO: organic to lead”.
2. Import the Organic Search sessions segment and add it to Segment comparisons.
3. Click the pencil next to Steps. Set Step 1 to the event session_start, so every organic session enters the funnel.
4. Add Step 2: the event page_view with a parameter condition, page_location contains /services/ (or whichever page type sits in the middle of your journey).
5. Add Step 3: the page that holds your form or checkout. For example, page_view where page_location contains /contact/.
6. Add Step 4: your key event, such as generate_lead or purchase.
7. Click Apply. Leave Make open funnel off for now, so users must enter at Step 1.
8. Drag Device category into Breakdown to see if mobile and desktop organic visitors convert differently.

A funnel can have up to 10 steps, but three to five is easier to act on. Between steps you can choose is indirectly followed by (any pages in between) or is directly followed by (the very next step). I use indirect as the default for SEO, because visitors rarely take a straight line.

Useful Settings

• Show elapsed time shows the average time between steps. Long gaps on a contact page often point to a confusing form.
• Trended funnel, chosen under Visualization, plots each step over time. It’s useful after a core update or a page redesign.
• Within sets a time limit between steps, such as 30 minutes. Use it if you only care about conversions in the same visit.

Click a step in the chart to create a segment of the users who left at that point. You can then look at what those users did in a free-form exploration.

5. Cohort and Retention of Organic Users

Organic traffic is often judged on the first visit alone. I think that’s a mistake. A cohort exploration (a view that groups users by when they first arrived) shows if people who found you through search come back. That matters a lot for content sites, SaaS and anyone with a long sales cycle.

Build Steps

1. Go to Explore > Cohort exploration and name it “SEO: organic retention”.
2. Set Cohort inclusion to First touch (acquisition date). This groups users by when they first arrived.
3. Set Return criteria to Any event for general retention. Or set it to a key event if you want to measure returning conversions.
4. Set Cohort granularity to Weekly. Daily is too noisy for most sites, and Monthly hides changes.
5. Leave Calculation on Standard, which counts users who came back in each period.
6. Import the dimension First user default channel group and drag it into Breakdown. Each cohort now splits into Organic Search, Direct, Referral and the rest.
7. Under Values, keep Active users, or switch to a key event count or Total revenue.

Use a breakdown here, not the organic session segment. The question is about users who first came through search. First user default channel group is the user-level dimension that records exactly that. The breakdown also gives you other channels to compare against.

What It Tells You

• If organic users come back at a lower rate than direct or email users, your content may answer one-off questions without giving people a reason to return.
• A drop in week-one retention for recent cohorts can follow a change in which pages rank, even when total organic sessions stay flat.
• Switch Calculation to Cumulative with revenue as the metric. This shows the longer-term value of organic users, which helps when you have to justify SEO budget.

Thresholds, Sampling and Limits to Know

Explorations are more flexible than standard reports, and that comes with limits. Google lists them on its explorations help page.

Sampling

Explorations may be sampled (based on part of the data, not all of it) when a query covers more than 10 million events in a standard property. The data quality icon at the top right turns from green to yellow when sampling or thresholds apply. Hover over it to see why.

The simplest fix is a shorter date range. Run a few shorter periods and compare them, not one long query. Standard reports aren’t sampled, so I use them to double-check exploration totals.

Thresholds

GA4 hides rows when there are too few users, to protect privacy. This happens most often when you add demographic dimensions such as age or gender, or audiences built on demographics. You can’t turn thresholds off.

If rows go missing, remove demographic dimensions first. Then widen the date range. More users in the range often brings hidden rows back.

Other Limits

• Retention window: as covered above, explorations can’t go back further than your data retention setting.
• Exploration limits: each user can create up to 200 explorations per property, with up to 500 shared per property. Each exploration takes up to 10 segments, 20 dimensions and 20 metrics.
• Sharing: explorations are private until you click Share. Shared ones are read-only for others, so colleagues who want to edit must make a copy.
• Date ranges: preset ranges such as “Last 28 days” move forward when you reopen an exploration. Custom ranges stay fixed, which is handy for before-and-after comparisons.
• No Search Console data: queries, impressions and positions live in the Search Console reports collection, not in Explore. Use your explorations next to Search Console. Don’t expect one tool to answer everything.

GA4 also undercounts when visitors decline cookies or block tracking. That affects every channel. Keep it in mind when you compare GA4 sessions with Search Console clicks.

Turning Explorations Into a Monthly Routine

Once all five are built, reviewing them takes less than an hour a month. This is the order I follow.

1. Open the landing page exploration. Note the pages whose engagement or key event rate moved most.
2. Run the path exploration from one of those pages to see where visitors go next.
3. Check the organic vs AI comparison for pages gaining AI referrals.
4. Review the funnel for the biggest drop-off and pick one fix.
5. Glance at the latest cohorts to check that retention isn’t slipping.

Pages that keep their traffic but lose engagement are good picks for a content refresh.

Conclusion

Good GA4 explorations SEO work comes down to five views you can reuse: organic landing pages with results, the paths visitors take, organic compared with AI assistant traffic, a funnel to your key event, and how many organic users come back. Build the organic segment once, set retention to 14 months, and you can build all five in an afternoon.

I’d start with the landing page exploration, because it feeds the other four. If your key events aren’t reliable yet, fix those first. That way every exploration measures the actions that really matter.