Funnel Analysis on Google Analytics Data by Tatevik KhachatryanFunnel Analysis on Google Analytics Data by Tatevik Khachatryan

Funnel Analysis on Google Analytics Data

Tatevik Khachatryan

Completed work

Data Analyst

Dash Plotly

Google Analytics

Google BigQuery

E-Commerce

Like this project

Completed work

Posted May 20, 2025

Performed funnel analysis on GA data to identify conversion drop-offs and suggest improvements.

Likes

Views

Timeline

Apr 10, 2025 - Apr 20, 2025

Funnel Analysis on Google Analytics Sample Data

🧠 Project Overview

Two Types of Funnels: Standard vs Fast-Track

📱 Device-Level Funnel Drop-Off

🌍 Funnel Drop-Offs by Country

🌐 Drop-Offs by Browser

🧠 Project Overview

This project uses Google Analytics session data from BigQuery's public dataset GA_Sessions_Demo to perform a funnel analysis across key ecommerce steps:

Landing Page → Product Page → Cart → Checkout → Thank You

With over 10,000 sessions, this dataset offers a rich opportunity to identify conversion drop-offs, assess user behavior by segment, and suggest UX and technical improvements based on actual funnel performance.

Defining the Funnel

The raw GA data doesn't come with ready-made funnel steps, so I had to build them manually using the pagePath column. I tried two approaches:

Manual Keyword Mapping

I initially built a dictionary mapping funnel steps to specific pagePath patterns (e.g., Cart → /basket.html, Thank You → /ordercompleted.html).

However, this approach lost over half of the sessions due to overly narrow matching.

Smarter Path Classification

To fix this, I implemented a classifier function using regex and keyword heuristics.

It labeled each pagePath into funnel steps, then aggregated at the session level.

def classify_page_path(path):
    if 'basket' in path or 'cart' in path:
        return 'Cart'
    elif 'checkout' in path or 'payment' in path or 'signin' in path or 'yourinfo' in path or 'revieworder' in path:
        return 'Checkout'
    elif 'ordercompleted' in path:
        return 'Thank You'
    elif any(x in path for x in ['apparel', 'bags', 'accessories', 'electronics', 'drinkware', 'notebooks']):
        return 'Product Page'
    elif any(x in path for x in ['home', 'store', 'search', 'quickview']):
        return 'Landing Page'
    else:
        return 'Other'
results['step_type'] = results['pagePath'].apply(classify_page_path)

This method captured significantly more sessions and improved funnel fidelity.

Two Types of Funnels: Standard vs Fast-Track

While analyzing, I noticed 114 sessions went straight to Checkout, skipping Cart entirely — likely due to quick-buy buttons or direct checkout links. So I split the funnel into two paths:

📉 Fast-Tracked Funnel Results

Fast-tracked sessions had 0 completions, suggesting:

Users dropped off mid-checkout

Something broke in the quick checkout flow Or possibly bot/misclassified sessions

Suggestion: Investigate fast-track checkout UX and session origins. Consider bot filtering.

📉 Standard Funnel Results

From 1574 standard sessions:

1086 reached a Product Page

252 reached the Cart (76% drop-off)

51 reached the Thank You page

That’s a 3.2% overall completion rate.

Suggestions:

Improve Cart transition UX (e.g., clearer CTAs, trust badges)

Simplify checkout forms & reduce steps

Run A/B tests on Product Page layout and CTA visibility

📱 Device-Level Funnel Drop-Off

Segmenting by deviceCategory revealed huge behavioral differences.

🔍 Insights:

Mobile conversion = 0.5%, vs Desktop = 4.2%

Tablet = 0% conversion

Suggestions:

Prioritize mobile UX and error tracking

Consider deprioritizing tablets unless traffic increases

🌍 Funnel Drop-Offs by Country

Using country-level segmentation, I found that while the US has the highest drop-off volume, international users like Germany, India, and the UK show higher drop-off rates, pointing to potential trust or localization issues.

Suggestions:

Run localized A/B tests on product + cart flow

Add country-specific shipping info and currency displays

🌐 Drop-Offs by Browser

I categorized 15+ browsers into High and Low volume groups.

Among High-volume browsers:

Chrome:

Largest drop-off volume across the funnel

654 users dropped from Product → Cart

191 dropped at Checkout → Thank You

Safari & Firefox:

Drop-off rates > 80% at both Cart and Checkout stages

Internet Explorer:

100% abandonment at Checkout

Likely poor compatibility or broken UX

Common Drop-Off Issues:

Tracking gaps (negative drop-offs → session stitching issues)

Checkout flow inconsistencies

Cart not tracked properly

Suggestions:

QA checkout pages in Firefox, Safari, and IE

Improve session stitching and tag firing reliability

Final Takeaways

✅ Mapped a 5-step ecommerce funnel across 10K+ GA sessions

✅ Identified drop-offs by segment: device, country, browser

✅ Created actionable insights for product, UX, and data teams

✅ Built two funnel models: Standard and Fast-Track

✅ Used BigQuery + Colab for data processing and Plotly for visualization

Got thoughts, feedback, or just want to say hi? Drop me a “Hi 👋” on LinkedIn.