How to Use Google Analytics 4 Explorations: 5 Reports Every Marketer Should Build

by | Nov 22, 2026 | 0 comments

In short: GA4 explorations live in the Explore section of Google Analytics 4 (GA4). There you build your own exploration reports from raw event data instead of reading a fixed report.

Five techniques answer most marketing questions: free form, funnel, path, segment overlap and cohort. Each one below comes with the exact dimensions and metrics to drag in, and the decision it helps you make. Check one setting before you trust any of it: data retention, which starts at two months.

What explorations are, and why their numbers behave differently

Standard reports in the Reports section answer the same questions for every property, through the same interface: where traffic comes from, which pages get views, which events fire. Explorations answer your own question. You pick the technique, the dimensions, the metrics, the segments and the filters, and GA4 builds the table or chart on the fly.

The difference comes from where the data lives. Standard reports read aggregated tables that Google prepares in advance and keeps for the life of the property. Explorations query the raw data, event by event, because nobody knows in advance which combination you’ll ask for. That one design choice explains the three limits that catch most marketers out:

  • Retention: explorations only reach back as far as your event data retention setting, which is 2 or 14 months on a standard property.
  • Sampling: a very large query can be computed on a subset of events. Google’s help center puts the limit at 10 million events per query for standard properties.
  • Thresholds: when Google signals or demographic data are involved, GA4 can hide rows with few users to protect their privacy.

It also explains why a figure in an exploration rarely matches the same figure in a standard report. Neither one is wrong. They come from different tables, and sometimes from different counting rules, so compare trends rather than exact totals.

analytics dashboard laptop

Prepare the property before you build anything

Ten minutes of setup saves hours of confused analysis. Start with retention, because it can’t be fixed after the fact. Go to Admin > Data collection and modification > Data retention and set event data retention to 14 months. The change only applies from now on: data older than your current setting is already gone.

Next, check that the actions you care about are marked as key events. Google renamed conversions to key events in 2024, so older tutorials still say Conversions. For a shop, the recommended ecommerce events are view_item, add_to_cart, begin_checkout and purchase. For lead generation, generate_lead is the recommended event name, and enhanced measurement can record form_start and form_submit.

Then exclude your own team. Define internal traffic in the web data stream’s tag settings, and activate the matching data filter in Admin.

Finally, keep consent in mind. Under the GDPR (the EU’s General Data Protection Regulation) and the ePrivacy rules, analytics cookies need prior consent in the EU. Visitors who refuse are partly or fully missing, which is one more reason to read explorations as trends.

Inside the Explore section: variables, tab settings and canvas

Click Explore in the left menu. You’ll see your saved explorations, a Blank option and a link to the Template gallery. The gallery sorts ready-made explorations into three families: techniques, use cases and industries. Exploration templates save setup time, but a blank exploration teaches you how each piece fits together.

Every exploration has three areas:

  • Variables (left column): the exploration name, date range, segments, dimensions and metrics. Everything here is available to every tab.
  • Tab settings (middle column): the technique, and where each variable goes, such as rows, columns, values, breakdowns or filters. Changes here affect the current tab only.
  • Canvas (right): the result. One exploration can hold several tabs, each with its own technique.

Building is drag and drop. Use the plus sign next to Dimensions or Metrics to import a field, then drag it into Rows, Values or Breakdown. Use the plus sign next to Segments to create a segment that any tab can reuse.

A greyed-out field is incompatible with the technique or with another field already in place. For example, an item-scoped dimension such as Item name doesn’t combine with an event-scoped metric such as Total revenue, so pair it with Item revenue instead.

Report 1: the channel and landing page table (free form)

The question: which traffic source, channel and landing page bring visits that turn into results? Free form is the default technique and the most flexible one. It offers six data visualization types: a table by default, plus a donut chart, line chart, scatterplot, bar chart or geo map.

  1. Choose Free form and name it Channel and landing page performance.
  2. Import the dimensions Session default channel group, Landing page + query string and Month.
  3. Import the metrics Sessions, Engaged sessions, Engagement rate, Key events and Session key event rate.
  4. Drag the channel group, then the landing page, into Rows, and set Nested rows to Yes.
  5. Drag Month into Columns and the five metrics into Values.
  6. Raise Show rows so smaller landing pages appear too.

Session key event rate is the conversion rate under its GA4 name: the share of sessions with at least one key event. Engagement rate is engaged sessions divided by sessions. According to Google Analytics Help, a session counts as engaged when it lasts longer than 10 seconds, includes a key event, or has at least 2 page or screen views. You can raise that 10-second timer in the web stream’s tag settings.

The decision: a landing page with many sessions and few key events brings traffic and then loses it. Check whether it matches the search intent and whether the call to action is visible. A page with modest traffic and a strong session key event rate deserves more internal links and more promotion. For channels, compare session key event rate before you move budget, never sessions alone.

For SEO (search engine optimization) work, add a filter where Session default channel group exactly matches Organic Search. GA4 shows what happens after the click, while Google Search Console still owns queries, impressions and rankings.

analytics dashboard laptop

Report 2: where people leave your journey (funnel exploration)

The question: at which step of a journey you designed do people give up? A funnel exploration takes up to 10 steps, and each step is a condition on an event or a page.

  1. Choose Funnel exploration and open the steps editor in Tab settings.
  2. For a shop, create four steps: view_item, add_to_cart, begin_checkout, purchase.
  3. For lead generation, use three: a page_view where Page path contains /services/, then form_start, then generate_lead.
  4. Drag Device category into Breakdown and switch on Show elapsed time.
  5. Add Page title as the Next action dimension to see what people do instead of the next step.
  6. Keep the Standard funnel view, and switch to Trended funnel to follow one step over time.

The funnel is closed by default: people must enter at step one. Toggle Make open funnel if visitors can legitimately enter halfway, for instance on a checkout reached from an email.

The decision: fix the step with the largest drop first, on the device where it happens. Example: if mobile users go from begin_checkout to purchase at half the desktop rate, test the mobile checkout before you buy more mobile traffic. The Next action column often shows the culprit, such as a shipping page or a return to the cart.

Report 3: the routes nobody planned (path exploration)

The question: what does the real user journey look like before people convert? A funnel tests the route you designed, whereas a path shows the routes you didn’t design. The team at infotrust.com reached a similar conclusion.

  1. Choose Path exploration, then click Start over at the top right.
  2. Pick Ending point rather than Starting point, and select the event name purchase or generate_lead.
  3. Set the node type to Page title and screen name, or Page path and screen class if your titles repeat.
  4. Switch on View unique nodes only so a page reloaded twice doesn’t fill a step.
  5. Drag Device category into Breakdown, and add an Organic Search segment if you want the SEO view.

Click a node to expand the next or previous step. Working backwards from the conversion keeps you focused on journeys that end well, instead of the thousands that don’t.

Path data also shows how user behavior differs by device: a mobile journey with more back-and-forth steps often points to a layout problem.

The decision: pages that keep showing up just before the conversion, such as pricing, case studies or an About page, are doing persuasion work. Link to them from your top landing pages. Loops between two pages usually signal missing information. A second tab can start from the homepage instead: heavy use of site search right after landing often means the navigation hides what people came for.

Report 4: who your best customers really are (segment overlap)

The question: how do your audiences intersect, and which intersection buys? Segment overlap compares up to three different segments in a diagram, with a table underneath.

  1. Choose Segment overlap.
  2. Create three user segments with the plus sign next to Segments:
    • Mobile users: Device category exactly matches mobile.
    • Organic users: First user default channel group exactly matches Organic Search.
    • Purchasers (event purchase) or Leads (event generate_lead).
  3. Drag Country or Session default channel group into Breakdowns.
  4. Put Active users, Key events and Total revenue in Values.

Right-click an intersection to create a segment from it, or to view the number of users behind it. In the segment builder, the Build an audience option turns that segment into an audience you can reuse in Google Ads.

The decision: if the Organic and Purchasers circles overlap heavily, organic search feeds revenue, so defend the pages that rank. If buyers sit almost entirely outside the mobile circle, check the mobile experience before you cut mobile spend. The answer may be friction rather than a lack of intent.

analytics dashboard laptop

Report 5: do customers come back? (cohort exploration)

The question: do people acquired through one channel return and buy again? A cohort is a group of users who share a characteristic, here the week they were acquired.

  1. Choose Cohort exploration.
  2. Set Cohort inclusion to First touch (acquisition date).
  3. Set Return criteria to Any transaction for a shop, or to your lead event for a service business.
  4. Set Cohort granularity to Weekly and Calculation to Standard.
  5. Drag First user default channel group into Breakdown, and use Active users or Transactions as the metric.

Switch to Monthly granularity for B2B (business-to-business) sales with long cycles, and to Cumulative when you want the running total of repeat buyers. With only two months of retention, you’ll see about eight weekly cohorts at most, which is another reason to set 14 months.

The decision: a channel that brings fewer first buyers but a better return rate deserves more weight than its first-purchase results suggest. For a content site, compare how often readers who arrived on a blog post from the newsletter return against social visitors before you decide where to invest your writing time.

Two more techniques: user explorer and user lifetime

User explorer lists pseudonymous IDs and lets you open the event stream of one individual user. It gives the most granular view GA4 offers, and it’s handy when you audit your tracking, for instance that a purchase fires once and not twice. Never send names, emails or phone numbers to GA4, because Google’s policies forbid personally identifiable information there. See owox.com for their take.

User lifetime reports totals over each user’s whole history, such as lifetime revenue by first user source. It suits subscription and repeat-purchase businesses, which need to know which campaigns bring customers worth keeping.

Which exploration to build first

Start with the report that matches your business model, then add the second one once the first has changed a decision.

If you run Build first Then Avoid
An online shop Funnel from view_item to purchase, by device Cohort on Any transaction A path from every page with no ending point: too much noise
A lead generation or B2B site Free form by channel and landing page, with session key event rate Path ending on generate_lead Daily cohorts: too few leads per cell, and thresholds hide rows
A content site or blog Free form on landing pages with engagement rate Path from the top landing pages Long funnels: almost nobody completes six steps
An agency reporting to clients A saved free form table, plus a Looker Studio dashboard for the client Segment overlap for audience ideas Sending explorations as the monthly report
analytics dashboard laptop

Sharing, exporting and the limits of explorations as reports

Explorations are private until you share them. Click the share icon at the top right to share one with people who already have access to the property. They get a read-only copy and can duplicate it to make changes. The export icon sends the data to Google Sheets, or downloads it as a CSV (comma-separated values) file or a PDF. idatafusion.com walks through the specifics.

You can’t pin an exploration to the Reports section. To get a similar view there, someone with the Editor role can build a custom detail report from Reports > Library.

For client dashboards, Looker Studio is the better fit. It uses your GA4 property as a native data source, and you can share a report through a link that doesn’t require a login. Keep in mind that viewers then see the data through the owner’s credentials.

Explorations don’t replace attribution either: credit across channels lives in the Advertising section, where data-driven attribution is the default model. In our view, agencies shouldn’t build monthly client reporting on explorations. Retention caps history at 14 months, shared copies are read-only, and every viewer needs property access. If you need year-over-year history beyond that, turn on the free BigQuery export early, because it only stores data from the day you link it.

Five pitfalls, and how to fix them

  • The funnel doesn’t match the Pages and screens report. Cause: the funnel counts users who meet each step in order, while the page report counts every view. Fix: build steps on events, compare an open funnel with the report, and judge trends rather than totals.
  • The date picker won’t go back further than two months. Cause: the default retention setting. Fix: switch to 14 months today, and use standard reports or BigQuery for older data.
  • Rows vanish or totals don’t add up, with a thresholding notice. Cause: data thresholds linked to Google signals or demographics. Fix: widen the date range, drop demographic dimensions, or set Reporting identity to Device-based in Admin.
  • The data quality icon warns about sampling. Cause: the query exceeds the event quota. Fix: shorten the date range, remove a breakdown, or move the analysis to BigQuery.
  • The exploration shows your own visits. Cause: the internal traffic filter is defined but still in Testing mode. Fix: set it to Active, knowing it only affects data collected from then on.

How to know your explorations are paying off

An exploration earns its place when it changes a decision, so measure the decision rather than the report. After you rework a landing page found in report 1, watch its session key event rate for four weeks. After a checkout change, watch the completion rate of the step you fixed in report 2.

A workable rhythm is the free form table monthly, the funnel weekly while you test, cohorts monthly, and paths once a quarter. Keep a simple change log with dates, because otherwise you’ll forget which change moved which curve. Correlation isn’t causation: seasonality, tracking changes and a new marketing campaign can all move a number at the same moment.

Checklist: your first five GA4 explorations

  1. Set event data retention to 14 months.
  2. Mark the right key events and activate the internal traffic filter.
  3. Build the free form table by channel and landing page.
  4. Build the funnel from first step to key event, broken down by device.
  5. Build the path ending on your main key event.
  6. Build segment overlap with mobile, organic and buyers.
  7. Build a weekly cohort by first user channel.
  8. Check the data quality icon on every tab.
  9. Share read-only copies with the team, and use Looker Studio for clients.
  10. Write down the decision each report supports, and review it monthly.

Pick the one question that costs you the most money today, whether it’s checkout drop-off, weak landing pages or one-time buyers, and build only that exploration this week. Once it has changed a decision, the other four will take far less time to build. For more practical guides on analytics and digital marketing, browse the rest of our blog.

FAQ

How do I find AI traffic in GA4?

Build a free form exploration with Session source as rows and Sessions and Key events as values. Add a filter on Session source that matches a regular expression such as chatgpt|perplexity|gemini|copilot|claude. Treat the result as a minimum: a visit only shows its source when the browser passes a referrer, and GA4 files the rest under Direct.

How is GA4 different from Universal Analytics?

Universal Analytics was built around sessions and pageviews, while GA4 records everything as events with parameters. Standard Universal Analytics properties stopped processing data on July 1, 2023. Explorations replace the old custom reports, with far more techniques.

Do explorations change my data or what my colleagues see?

No. An exploration only reads data, and it stays private until you share it. Deleting one removes the analysis, not a single event.

Are explorations worth it on a small site?

Yes, but with fewer techniques. With a few hundred sessions a month, thresholds and tiny counts make cohorts and segment overlap hard to read. A free form table over three months and a short funnel still give useful answers.

Do I need Google Analytics 360 to use explorations?

No. All the techniques in this guide are available in the free version of GA4. The paid version mainly raises limits such as sampling quotas and data retention.

Search

Recent Posts

Subscribe now