eCommerce Funnel Analysis: Finding the Real Leaks

Every store has a funnel, and every funnel leaks. The trouble is that the leak you can see is rarely the one costing you the most. Teams obsess over a 70% cart abandonment rate while ignoring that nine in ten visitors never reach a product page at all. Funnel analysis done properly isn’t about staring at a tidy descending bar chart — it’s about quantifying each leak in revenue terms so you can fix the one that actually matters first.

Map the funnel before you measure it

A generic five-step funnel — visit, product view, add to cart, checkout, purchase — is a starting point, not your funnel. Real shopper journeys branch: people search, filter, compare, leave, and return. Before measuring, map the steps that genuinely reflect how people buy from you.

A workable structure for most stores:

1. Landing — the visitor arrives
2. Engagement — they view a category, search, or product
3. Product consideration — they view a product detail page
4. Intent — they add to cart
5. Checkout — they begin the checkout flow
6. Purchase — they complete

The right granularity matters. Too coarse and you can’t locate the leak; too fine and you drown in noise. Add a step only when a meaningful share of shoppers drop off there.

Quantify each leak — in revenue, not percentages

Conversion percentages mislead because they ignore volume. A step with a scary drop-off rate may touch few shoppers, while a modest drop on a high-traffic step bleeds far more revenue.

For each step, calculate:

• Drop-off rate — the share who don’t proceed
• Volume — how many shoppers reach the step
• Revenue at risk — drop-off volume multiplied by the expected value of a shopper at that stage

That third figure is what turns analysis into prioritisation. A 5% improvement at a step that 100,000 shoppers reach will almost always beat a 30% improvement at a step only 2,000 reach. We always frame funnel work in revenue at risk, because it forces honest comparison between leaks of different shapes. It’s the same logic behind focusing on revenue per visitor rather than conversion rate alone.

Segment, or you’ll fix the wrong thing

An aggregate funnel is an average of very different journeys, and averages hide the truth. The single most valuable move in funnel analysis is to break it down by segment.

By device

Mobile and desktop funnels almost always diverge, usually with mobile losing more shoppers at checkout. If your aggregate funnel looks “fine”, a mobile conversion gap may be hiding inside it.

By traffic source

Paid, organic, email, and direct traffic arrive with different intent. A leak concentrated in one source points to a mismatch between the promise that brought them and what they found.

By new versus returning

Returning shoppers convert differently and leak at different steps. Mixing them obscures both.

By landing context

Shoppers who land on a category behave differently from those who land on a product or a search result. The leak often lives in one entry path, not the whole funnel.

When you segment, the same funnel frequently reveals that your “checkout problem” is really a “mobile checkout problem” or a “paid-traffic relevance problem” — a far more actionable conclusion.

Diagnose the why behind each leak

A drop-off tells you where, not why. To find the cause, layer qualitative evidence onto the quantitative leak:

• Session recordings of shoppers who dropped at the step
• Form analytics for checkout and account leaks — which field causes hesitation or errors
• On-page surveys triggered at the leak (“What stopped you from continuing?”)
• Support tickets and reviews mentioning the same friction

Common culprits by step:

• Landing → engagement: slow load, mismatch with the ad or query, weak above-the-fold relevance
• Engagement → product: poor category navigation or weak site search, which is often a store’s highest-intent surface and the easiest place to lose motivated shoppers
• Product → cart: unclear pricing, thin product information, sizing or stock uncertainty
• Cart → checkout: surprise shipping costs, forced account creation
• Checkout → purchase: payment friction, errors, lack of trust signals

Turn the analysis into a prioritised plan

Funnel analysis only earns its keep when it produces a ranked action list. For each significant leak:

1. State the leak and its revenue at risk.
2. Note the segment where it concentrates.
3. Record the likely cause from your qualitative evidence.
4. Define a hypothesis and a test.
5. Estimate effort, and rank by impact-over-effort.

Then test the changes rather than assuming the fix works — funnel analysis identifies the target; experimentation confirms the cure. Our guides to AI-powered CRO and reducing cart abandonment cover how to run those experiments and the specific fixes for the checkout end of the funnel.

Common pitfalls

• Reading the aggregate funnel only. The real leak almost always hides inside a segment.
• Optimising the scariest percentage instead of the largest revenue leak.
• Treating the funnel as linear. Shoppers loop and return; build for the actual journey.
• Stopping at “where”. Without the “why”, you’re guessing at fixes.
• One-off analysis. Funnels shift with seasonality, traffic mix, and site changes. Review regularly.
• Ignoring micro-conversions that don’t pay. Lifting add-to-cart means nothing if those carts never convert.

Where to start

Build one clean funnel this week, broken down by device and traffic source. Calculate revenue at risk for each step. You’ll almost certainly find that the leak draining the most money isn’t the one your team has been worrying about. Fix that one first, measure the change, and move to the next.

Funnel analysis is the cheapest high-leverage work in eCommerce: it costs analysis time, not engineering, and it tells you exactly where to point everything that follows. If you’d like help mapping your funnel and quantifying where the revenue is leaking, our conversion optimisation work starts here. Get in touch and we’ll find your biggest leak together.