GA4 Exploration Reports — Funnel and Cohort Analysis for Customer Journey

You have Google Analytics 4 set up, you check Pages and screens report, but you can't figure out where customers drop off. Traffic is up, sales are not. Standard reports tell you what happened, not where and why users abandon. You need a scalpel, not a hammer.

We, from Meteora Web, work daily with GA4 for our clients — e-commerce, SaaS, small businesses in Southern Italy. Exploration reports are the least used yet most powerful part of the tool. Path funnels and cohort analysis let you see exactly at which step the user stops and whether your recent investments bring quality customers or just one-time clicks.

This guide starts from the real problem: you have data but can't read it. We'll show you how to build a funnel report to identify bottlenecks and a cohort report to understand if users come back. No theory: real steps, numbers, actions you can take right now.

Why standard GA4 reports are not enough to spot conversion leaks

The Explorations tab is the least used part of GA4. Google hid advanced features behind a different interface. But this is where the truth lies.

A clothing e‑commerce client came to us saying: “I have 10,000 visitors per month but only 50 orders. The site looks good, what's wrong?” We opened a funnel report and discovered 40% of users left at “Add to cart” because the button was hidden on mobile. Standard reports didn't show that. The funnel did.

Funnel concept: not a linear pipeline

In GA4 the path funnel does not assume all users follow the same sequence. You can define open steps (user may skip) or sequential steps (must pass through all). We always use open steps for e‑commerce: a user can reach the product page without visiting the homepage.

The difference between funnels and standard reports is that the funnel tells you the drop‑off percentage between each step, not just how many people viewed a page.

How to create a path funnel report in GA4 step by step

Open GA4, go to Explorations (left menu, under Reports). Click “Blank” or select the “Path funnel” template. We recommend starting with “Blank” then selecting “Path funnel” as technique.

Define funnel steps

Imagine a typical purchase path for an e‑commerce product:

• Step 1: view product page
• Step 2: add to cart
• Step 3: begin checkout
• Step 4: purchase completed

In GA4 each step is an event. Default names are: view_item, add_to_cart, begin_checkout, purchase. If you have custom events, use those.

Configure the Path funnel technique

In the exploration, select “Technique: Path funnel”. Add segments: “Step 1: event = view_item”, “Step 2: event = add_to_cart”, etc. Set the time window to 7 days (users who complete the funnel within 7 days).

Here is a real example:

Step 1: view_item (product page)
Step 2: add_to_cart
Step 3: begin_checkout
Step 4: purchase

Settings:
- Window: 7 days
- Include all users
- Options: “Sequential steps” (order matters)
- Show drop‑offs between steps: YES

After creating the report, you'll see a funnel chart with drop‑off percentages. In our experience, if 60% of users drop between view_item and add_to_cart, the problem is on the product page (price, images, CTA). If the drop is between add_to_cart and begin_checkout, there is likely an obstacle in the cart (unclear shipping, forced registration).

Interpret the numbers

Do not only look at the final conversion rate (e.g., 1%). Look at where you lose the most users in absolute terms. Sometimes a small improvement in an intermediate step yields more total conversions than optimising the last step.

We, from Meteora Web, saw a client increase from 1.2% to 2.8% conversion simply by moving the “Add to cart” button above the fold on mobile. The funnel report had highlighted a 70% drop on mobile between view_item and add_to_cart.

Cohort reports in GA4 — how to understand if users really come back

A funnel tells you where users get lost in a single session. A cohort tells you whether users acquired in a certain period return in subsequent weeks. It's the thermometer of retention.

Cohort concept: groups of users over time

A cohort is a group of users who performed a specific action in a period (e.g., first visit in January). Then you measure how many of them come back to do something (e.g., purchase) in the next month, two months later, etc.

If you invest in Google Ads in March and see that after one week only 5% of users return to the site, you are bringing low‑quality traffic. If a Facebook campaign has a 30‑day retention of 20%, that campaign is more effective in the long run.

How to create a cohort report in GA4 step by step

Go to Explorations, click “Blank”, select “Technique: Free form”, then switch to “Cohort”. Or use the “Cohort” template.

Set cohort dimensions

Choose:

• Cohort dimension: “First visit date” (or “Acquisition date” – but first visit is more precise for new user behavior).
• Cohort period: Weekly (good balance between detail and data volume).
• Include cohorts: Last 4‑8 weeks.
• Cohort metric: “Users” (to see how many return) or “Purchases” for monetary retention.
• Return interval: Days or weeks after the initial cohort.

Practical example:

- Cohort based on: First visit date (weekly)
- Cohort period: Weekly
- Return interval: 1, 2, 3, 4 weeks
- Metric: Users
- Filter: Only Google Ads traffic

The report will display a grid: each row is a cohort (e.g., users who arrived in week 1 of May), each column is the following week. The number shows how many users from that cohort returned. Percentage is usually shown as a retention rate.

Read the report: where to act

If retention at 2 weeks is below 10%, you are probably acquiring users with wrong expectations or the site does not offer enough value for a return. Concrete actions:

• Improve post‑purchase communication (email, remarketing).
• Offer useful content (blog, guides) to attract non‑paid return visits.
• Segment cohorts by source: compare organic vs paid retention. Organic often has higher retention.

Funnel vs Cohort — when to use each

Funnel: analyses a single flow, typically within a session or a few days. Useful for optimising the conversion process.

Cohort: analyses behaviour over time, to understand retention and long‑term value (LTV).

Use them together: first identify the bottleneck with the funnel, then verify whether users who pass that bottleneck return (cohort). If after improving the cart the 30‑day retention stays low, the problem lies upstream.

Common mistakes in GA4 exploration reports

• Not using filters: If you don't filter by device or source, data is aggregated and you miss differences. Always filter by mobile/desktop or campaign.
• Too narrow time window: A 1‑day window loses users who complete the funnel over multiple days. Use 7 or 14 days.
• Ignoring custom events: If you don't have events for key steps (e.g., “Start registration”, “Complete form”), the report is blind. Set them up first.
• Cohort with too few data: Fewer than 1000 users per cohort makes the report noisy. Aggregate over longer periods.

We, from Meteora Web, have a rule: every client with significant traffic must have at least one funnel report and one cohort report active. On e‑commerce clients, the funnel always reveals at least one improvement opportunity worth 20‑30% more conversions.

If you want to dive deeper into the complete GA4 ecosystem, we have written a complete pillar guide on GA4 that covers everything from setup to advanced reports.

What to do now

1. Create a funnel report for your main conversion path (purchase, signup, lead). Set at least 4 steps and analyse the drop between each.
2. Build a cohort report with weekly period and “Users” metric. Compare 30‑day retention across different sources.
3. Act on the first bottleneck: if drop is high between view_item and add_to_cart, work on the product page. If between add_to_cart and begin_checkout, review the cart.
4. Repeat the cycle: after implementing a change, wait at least 2 weeks and check if the funnel improves. Don't stop at the first attempt.

The difference between those who use GA4 and those who master it lies in exploration reports. The data is already there. It's up to you to extract it.