How to use Drop-Off Analysis
Drop-off analysis tells you which funnel stage is bleeding the most users — so your next experiment fixes the actual leak, not the one your gut guessed at.
Drop-Off Analysis
The practice of measuring which funnel stage loses the highest percentage of users, so you fix the largest leak first.
Drop-off analysis is the diagnostic step that turns raw funnel data into a prioritised fix list. You walk every defined stage — landing, product, cart, checkout, payment, thank-you — and compute the percentage of users who entered the step but never reached the next one. The stage with the steepest fall is your starting point.
It sits underneath funnel analytics as the question that actually drives action. A funnel report shows you the shape; drop-off analysis tells you where to spend the next two weeks of engineering and test design. Done well, it kills the recurring CRO debate about which page 'feels broken' and replaces it with one number per step.
Most CRO programs lose months testing the wrong page. A team will redesign the product detail page because it 'looks tired' while 71% of the real loss is happening on a broken shipping calculator three steps later. Drop-off analysis exists to stop that pattern.
The mechanics are simple: define the stages, count users at each one, divide. The judgement work is in stage definition, segmentation, and knowing which drop is a real leak versus a stage that's supposed to filter people out.
What counts as a drop-off — and what doesn't
A drop-off is any user who entered a funnel stage and never fired the event that defines the next stage, within a session window you decide in advance (24 hours is typical for e-commerce, 7 days for considered purchases over €200).
Not every drop is a problem. Product-page exits from a 'New In' collection are mostly browsing behaviour and will always sit around 60-70%. Checkout-step exits, by contrast, are paid-in users abandoning a stated intent — that's where your money goes to die.
The trick is to score each drop against a benchmark for that stage, not against zero. A 45% cart-to-checkout drop sounds awful in isolation; it's actually slightly better than average for apparel. A 12% payment-to-confirmation drop sounds fine; it's catastrophic and almost always a payment-method or 3DS issue.
Don't analyse without segmenting
A single blended funnel hides the answer. Mobile vs desktop, paid vs organic, new vs returning, and geography (especially when Shopify Markets routes EU traffic through a localised checkout) each have different drop signatures. Always split at minimum by device and traffic source before you draw conclusions.
How to run drop-off analysis on a typical store
Start by writing down the canonical funnel for your store. For a Shopify apparel brand, that's usually six stages: landing, collection or PDP, add-to-cart, checkout-start, shipping/payment, and order-complete. Don't skip stages because you think they're 'always fine' — that's where leaks hide.
Pull at least 30 days of data so you average over weekly seasonality, and longer if you ran a campaign that skews the mix. Then compute the stage-to-stage conversion and its inverse, the drop-off rate, for each step. Sort descending. Your top one or two stages are the only ones worth a test sprint.
Drop-off by stage — typical Shopify apparel funnel
PDP-to-cart looks like the biggest leak by raw percentage, but that's expected behaviour — most visitors browse and leave. The interesting signals are the cart-to-checkout drop (a friction step worth testing) and any payment-step number above 10%, which usually points at a technical issue rather than a UX one.
Benchmark drops by vertical
Stage benchmarks vary sharply by vertical. Beauty and supplements convert harder from PDP to cart because of repeat buyers and subscriptions. Electronics carries longer consideration windows and worse cart-to-checkout numbers. Use the table below as a rough floor — anything noticeably worse is your priority.
Compare your numbers against the row that matches your category, not against an aggregate. An 80% PDP-to-cart drop is a problem for beauty and routine for furniture.
Typical stage-level drop-off rates by vertical (median ranges)
| Vertical | PDP → Cart | Cart → Checkout | Checkout → Payment | Payment → Confirmation |
|---|---|---|---|---|
| Apparel | 85-90% | 40-50% | 35-45% | 5-10% |
| Beauty & supplements | 78-85% | 30-40% | 25-35% | 4-8% |
| Electronics | 90-94% | 50-60% | 40-50% | 8-12% |
| Home & furniture | 92-96% | 55-65% | 45-55% | 10-15% |
| Food & beverage | 75-82% | 25-35% | 20-30% | 3-7% |
If your payment-to-confirmation number sits above the upper bound for your vertical, treat it as an incident, not an optimisation. It almost always means a card-network failure, a 3DS challenge that's silently rejecting users, or a Klarna/PayPal integration timing out. Engineering, not CRO.
Pitfalls that hide the real leak
The most common mistake is analysing a blended funnel and concluding 'checkout is fine'. Split the same data by mobile vs desktop and you'll often find mobile checkout converts at half the desktop rate — the desktop traffic was hiding it.
The second is misattributing exit cause. A user dropping at the shipping step might be leaving because of the price shown, not the UX of the step itself. Pair drop-off numbers with session replays or exit-survey data on the suspect stage before you brief a redesign.
Drop-off analysis is the start, not the answer
Finding the leakiest stage tells you where to look. It doesn't tell you why users leave. Pair the quantitative drop with one qualitative source — heatmaps, replays, or a short on-exit survey — before you write the hypothesis. Skipping that step is how teams ship redesigns that move nothing.
Drop-off analysis FAQ
Funnel analytics is the broader discipline of measuring user progression through defined stages. Drop-off analysis is one specific question inside it: which stage loses the highest percentage of users? You can't do drop-off analysis without a funnel, but you can build funnel reports for many purposes other than finding leaks.
For e-commerce, 24 hours from session start is the default — users who come back two days later are usually a new intent. For considered purchases over €200 or B2B-style flows, extend to 7 days. The window matters less than picking one and staying consistent across reports.
At least 30 days so weekly seasonality averages out, and ideally a period without an active promotion or paid-traffic surge that skews the channel mix. If a single stage has fewer than ~500 entries in your window, the percentage is too noisy to act on — widen the window or merge segments.
Look at the biggest gap versus the vertical benchmark for that stage, not the biggest absolute drop. A 90% PDP-to-cart drop is normal; a 15% payment-to-confirmation drop is a fire. Benchmarks tell you which numbers are anomalies and which are just the shape of the funnel.
Yes, but you analyse two funnels separately: the acquisition funnel (first purchase) and the retention funnel (renewal, reorder, upgrade). Blending them hides everything because returning customers convert 3-5x better at every step and drown out the new-user signal.
On Shopify and WooCommerce, yes — the standard checkout events are already tracked by most analytics platforms. You only need engineering when your funnel includes custom interactions (configurators, quizzes, gated content) where the events don't exist out of the box.
Minimum cuts: device (mobile/desktop), traffic source (paid/organic/email/direct), and new vs returning. For larger stores, add geography and landing-page type. Each cut should be a separate funnel view — never average across them in the same chart.
For apparel, 35-50% of users who start checkout complete it. Beauty and supplements sit higher at 50-70%. Electronics and furniture sit lower at 25-40%. Anything materially below your vertical's range is signal that drop-off analysis can localise to a specific step.
Two safeguards. First, compare to a stage-specific benchmark, not zero. Second, pair the quantitative drop with one qualitative source — session replay, heatmap, or exit survey — before you brief a test. Quantitative finds where; qualitative finds why.
Treat it as a monthly health check, plus an ad-hoc rerun whenever you ship a checkout change, switch payment providers, or see a sudden revenue dip. The leak moves over time — fixing the checkout step often surfaces a new leak upstream at PDP that was previously masked.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.