How to Find Where Your Ecommerce Funnel Is Leaking

You have looked at the funnel report a hundred times. Sessions on the left, orders on the right, four boxes shrinking in between. The numbers are fine. They are also useless.

Because the question is not "what is my conversion rate." It is "which step is leaking, for which segment, and why did it get worse this month?"

That is a different question. And it is the one your funnel dashboard will not answer.

What a funnel report shows. What it hides.

A standard ecommerce funnel report has four to five steps, depending on the platform:

• Session
• Product view
• Add to cart
• Checkout started
• Order placed

Each step shows a count and a conversion rate. The chart shrinks left to right. You stare at it. The narrowest gap is the obvious one to fix. You spend a sprint on the checkout page.

Three problems with this approach.

First, the aggregate funnel hides segment behaviour. Mobile users might be dropping at add-to-cart while desktop users sail through. Returning visitors might be converting fine while paid traffic bounces. The aggregate hides both.

Second, conversion rate trends are noisy at the daily level. The "drop" you panicked about on Monday might just be normal week-to-week variation.

Third, the funnel report does not tell you what changed between the weeks where it worked and the weeks where it did not. That is the part you need.

Connect the data sources

Anna connects to GA4 (the canonical funnel source for most stores), Shopify (for the actual order data), and your ad platforms (for the traffic source data). Read-only OAuth on each. Under five minutes end to end.

Once connected, she can join sessions, events, and orders across systems. The funnel becomes a queryable dataset rather than a fixed dashboard.

The first prompt

The most useful prompt for funnel analysis is the segmented one. Not "what is my funnel," but "which segment of my funnel is dragging the average."

Try this:

"Compute the conversion rate from session to order for the last 30 days, broken down by device, traffic source, and new vs returning visitor. Flag any segment where conversion has dropped meaningfully vs the prior 30 days."

Anna pulls GA4 sessions and events, joins to Shopify orders, segments by the variables you asked for, and runs a statistical test on each segment to flag the ones where the drop is bigger than normal noise. She does not just hand you 14 segments and let you eyeball — she ranks them and tells you which two or three matter.

A typical answer looks like:

The aggregate conversion rate barely moved. But mobile-paid-social dropped 42%. That is the entire story. Your "funnel issue" is one specific traffic source on one specific device, and you would never see it from the headline number.

The where-it-leaks follow-up

Once you know which segment is dragging, the next question is where in the funnel they are dropping.

Ask:

"For the mobile-paid-social segment, show me the step-by-step funnel rates for the last 30 days vs the prior 30 days. Where is the gap biggest?"

Anna runs the four-step funnel for that specific segment and highlights the step where the conversion rate has moved most. The problem is rarely "the funnel got worse." The problem is usually "one step got worse, for one cohort, for a specific reason."

If the drop is at session-to-product-view, you have a landing-page mismatch — paid social ads are sending traffic to pages that do not match the ad creative. If the drop is at add-to-cart-to-checkout, you have a price-shock or a cart-experience problem. If the drop is at checkout-started-to-order, you have a payment or trust issue.

Different leaks, different fixes. The aggregate funnel cannot tell you which.

The cart abandonment question

Cart abandonment is the most common funnel question and the one most often mis-answered.

The reflexive interpretation: "70% of carts abandon, our checkout is broken." The honest interpretation requires knowing what your specific abandon rate looks like vs benchmark, what it has been historically, and which segment is responsible.

Paste:

"What is the cart abandonment rate this month vs the last 6 months? Break it down by order value, payment method, and traffic source."

Anna computes the rolling abandonment rate, segments it, and surfaces what is unusual. The answer might be "abandonment rate is unchanged on average, but high-value carts (over $200) are abandoning at twice the historical rate." That is a different problem than a generic "fix checkout" project — it points to shipping, payment options, or a confidence issue at higher cart values.

The new-vs-returning split

Always run the new-vs-returning split. It is the single most informative segmentation in any ecommerce funnel.

Returning visitors are converting at 3-5x the rate of new ones. Always. The question is whether your new-visitor conversion is healthy for your category and your acquisition mix.

Ask:

"Show me new vs returning visitor conversion for the last quarter. For new visitors, segment by traffic source and tell me which source is converting at a rate disproportionate to its spend."

This is the real ROI question. Channels that bring lots of new visitors who do not convert are expensive. Channels that bring fewer new visitors at higher conversion rates are usually the underrated ones.

What you stop doing

Once GA4, Shopify, and your ad platforms are connected, the Monday-morning funnel post-mortem changes shape. You stop:

• Comparing dashboard screenshots from three tools
• Eyeballing whether a 0.4-point conversion drop is meaningful
• Treating the aggregate funnel as the funnel
• Running A/B tests on the wrong step because you misread the dashboard

You start asking segment-level questions and getting segment-level answers, in plain English, with statistical backing.

One question to start

If you are going to ask only one funnel question this week, ask this:

"Which segment of my funnel is responsible for the biggest conversion drop in the last 30 days, and at which step did they drop?"

It is the question every CRO project should start with. Most start without asking it, which is why most CRO projects optimise the wrong thing.

Connect GA4 and Shopify. Paste the question. Try it at heyanna.studio.