Cross-channel funnel leak finder: where buyers drop off
Checks that purchase tracking works, then builds a step-by-step e-commerce funnel per channel from GA4 events, compares it with the previous 30 days and pinpoints the biggest leak. Analysis only.
When to use it
When sales or conversion rate dropped, or buyers don't complete purchase, and you need to find which funnel step and channel is leaking.
What you get
A per-channel funnel table for the last 30 days versus the previous 30, with flagged drops, the single biggest leak estimated in lost purchases, its likely cause and three checks to confirm it.
Does it change anything in my account?
No. The assistant only reads your data and reports back.
Works with
Google Ads, GA4, Search Console
Fill in before sending
[property]
Map my e-commerce funnel for the last 30 days and compare it with the previous 30 days. GA4 property: [property]. Do not touch any settings. This is analysis only.
1. Check tracking first. Use Google Ads conversion tracking health and the GA4 key event list to confirm that purchase is recording. If purchase shows zero or broken data, stop and tell me before analysing anything.
2. Pull the unified overview for Google Ads, Search Console and GA4 for the last 30 days, plus the GA4 overview KPIs.
3. From GA4, pull event counts split by event name and session source/medium (dimension event, second dimension source_medium). Do this once for the last 30 days and once for the last 60 days, and derive the previous 30 days as the 60-day figure minus the 30-day figure. The funnel steps are view_item, add_to_cart, begin_checkout and purchase. These do not need to be key events; total event counts are enough.
4. If a step does not appear at all, list it as missing and build the funnel from the steps that exist. If only purchase exists, fall back to sessions → purchase by source/medium and landing page, and tell me which ecommerce events I should add.
5. Build one table per channel (group source/medium into paid search, organic search, paid social, direct, referral, email, other). Show the count at each step and the step-to-step rate, for both periods.
6. Flag any step whose rate fell more than 20% versus the previous period. Also flag any channel whose rate at a step is more than 30% below the site average.
Finish with:
- The single biggest leak, estimated in lost purchases: for the flagged step, (site-average rate minus channel rate) × the channel's volume entering that step, carried through to purchase at the channel's own downstream rates. Show the calculation and call it an estimate.
- The most likely cause, chosen from traffic quality, product page, checkout or tracking, with the evidence from the data that points to it and what the data cannot prove.
- Three concrete checks to confirm the cause before we add budget.
A one-page brief mapping who advertises in the client's market, where, with what message, and where the client stands today.
When to use itWhen you are about to start with a new client and need to understand who they compete against in paid and organic search before the kickoff call.