What Funnel Drop-off Benchmarks Can and Cannot Tell You

What are funnel drop-off benchmarks?

Published funnel drop-off benchmarks are cross-company averages of how many users leave each step of a defined flow, reported as a percentage per step or across the whole sequence. Vendors build them by pooling event data from the sites on their platform, then publishing a median or mean per industry. The number that comes out describes that vendor's customer base during that window. It is not a target your funnel is failing to hit.

Before you read any of the figures below, fix your own definitions. Drop-off arithmetic and the difference between per-step and end-to-end loss are covered in the drop-off rate breakdown, and the method for building the funnel itself is covered in funnel analysis. This page only deals with the question of what an outside number is worth once you have your own.

Why are cross-industry funnel drop-off benchmarks weak evidence?

Cross-industry funnel drop-off benchmarks compare numbers that were not produced the same way, so a gap between your rate and the published average carries no information about your product. Three things differ between the companies inside any average. Step definitions differ: one company counts a signup at email submission, another counts it at email verification, and the two denominators are not the same population. Traffic mix differs: a funnel fed by branded search behaves differently from one fed by paid social, and the published average does not say which fed the sample. Sequencing rules differ: some tools count a user as having reached step three only if they passed step two in the same session, others allow any order and any window.

Contentsquare states this problem in its own data. Its 2026 Digital Experience Benchmark, built on 99 billion web and app sessions across 6,500 or more websites and nine industries between Q4 2024 and Q4 2025, reports conversion rate as a year-over-year change of minus 5.1%, not as an absolute rate you can compare yourself against. A publisher with 99 billion sessions choosing to publish deltas instead of levels is telling you how far levels travel.

Which funnel benchmarks have published methodology?

The table below lists the funnel and conversion figures where the publisher discloses enough about sample and method to make the number checkable. Every row was read from the publisher's own page.

Two of the Baymard rows carry no publication year on the page they appear on. That is worth saying out loud instead of laundering into a confident citation.

What does the 70.22% cart abandonment average actually cover?

Baymard Institute's 70.22% is an average of 50 different studies on ecommerce shopping cart abandonment, published on a page last updated September 22, 2025, with the underlying studies spanning 2006 to 2025. It is the strongest number in this category because Baymard shows its inputs and its count. It is also an average across two decades of very different web, which means it describes the long-run shape of cart loss and not the state of your checkout this quarter. The narrower metric, and how it differs from checkout abandonment, is worked through in the cart abandonment rate formula.

Baymard's own reason breakdown, taken from a survey of US online shoppers whose sample size the page does not disclose, puts 40% of abandonments on extra costs being too high, 20% on delivery being too slow, and 19% on not trusting the site with credit card information. Those percentages are more actionable than the headline rate, because a shipping cost surprise is something you can find in a recording and remove this week.

How much does a step definition change the number?

A step definition change moves a reported drop-off rate by double digits without any user behaving differently. Take two SaaS companies that both publish a 62.00% drop-off between signup and activation.

Company A defines signup as email submitted. 10,000 users submit an email, 3,800 activate.

(10,000 - 3,800) / 10,000 x 100 = 62.00%

Company B defines signup as email verified. 10,000 users submit an email, 6,500 verify, 2,470 of those activate.

(6,500 - 2,470) / 6,500 x 100 = 62.00%

Both report 62.00%. Measured from the same starting population of 10,000 email submissions, Company A loses 62.00% and Company B loses 75.30%. The published benchmark hides 13.3 points of real difference behind one identical figure, and no reader of the benchmark can see which definition produced it.

What should you compare your funnel against instead?

Compare your funnel against your own funnel from a prior period, with the step definitions frozen. A four-week baseline on your own data controls for the traffic mix, the step boundaries and the sequencing rules that external averages leave undefined, which turns a change in the number into evidence about a change you made. Set the baseline before you ship anything, then read each step against it and treat any movement larger than your normal week-to-week variance as a signal worth investigating.

External benchmarks still earn one job: sizing the prize. Baymard's finding that a large-scale ecommerce site can gain a 35% conversion increase from checkout usability work, based on 4,400 or more test participant sessions across 25 rounds, is an argument for funding a checkout project. It is not an argument that your checkout is broken. For turning a measured step loss into a shipped change, see marketing funnel optimization and the reading order in the funnel report guide.

How does Flowsery turn a drop-off number into a cause?

Flowsery pairs the funnel with the recordings of the users who left it. Its built-in web analytics covers funnels, goals, real-time traffic, revenue by source and user journeys, so the step where users leave and the sessions behind that step live in one product. Funnel analysis in Flowsery shows the loss; the session replay shows what the user hit.

Detection runs on every session. Flowsery finds rage clicks, dead clicks, JavaScript errors, drop-offs and broken flows automatically, groups matching sessions into a single issue, and ranks issues by how many users hit them. Each issue arrives in Slack, Linear or Jira with the replay and steps to reproduce attached, and tagging @flowsery in Slack opens a draft pull request in your GitHub repo. The signals behind those groupings are described in frustration signals.

Frequently Asked Questions

What is a good funnel drop-off rate?

There is no published cross-industry figure that answers this, because the studies that would define it do not share step definitions. The closest defensible anchor is Baymard Institute's 70.22% documented cart abandonment average, drawn from 50 studies between 2006 and 2025, which applies to ecommerce carts and to nothing else. For any other funnel, your own prior period is the only comparison that controls for how the steps were drawn.

Why do two analytics tools report different drop-off for the same funnel?

The tools apply different sequencing and attribution rules to the same events. One counts a user at step three only if step two happened first in the same session, another counts any order within a lookback window, and a third deduplicates by user while the second counts sessions. Read each tool's funnel settings before treating the gap as a data quality problem.

Is cart abandonment rate the same as funnel drop-off?

No. Cart abandonment rate is one specific end-to-end drop-off measurement in one specific flow, from cart creation to completed order. Funnel drop-off is the general case and can be measured at any step of any defined sequence. The two are calculated differently enough that the cart abandonment rate page treats them separately.

How large a sample do I need before my funnel numbers mean anything?

Enough users at the narrowest step that a handful of individuals cannot move the percentage. If your final step sees 40 users a week, a swing of four people is a 10 point move and tells you nothing. Widen the window until the smallest step carries a few hundred users, then read the trend instead of the single week.

Should I ever quote an industry benchmark to my team?

Quote it to size an opportunity, not to grade your funnel. Unbounce's 6.6% median landing page conversion rate, from 41,000 or more pages and 57 million conversions in 2024, is a fair way to argue that landing page work pays. It is not evidence that a specific page of yours is underperforming.

What do I do first when a step shows a large drop?

Watch the sessions that ended at that step before changing anything. A drop-off number tells you where users leave and never why, and the why is in the recording: a form field that rejects a valid input, a button that does nothing, a script error on a specific browser. Flowsery groups those sessions into one ranked issue so you read the pattern instead of ten individual replays.

Why does Contentsquare report a year-over-year change instead of an absolute conversion rate?

Contentsquare's 2026 Digital Experience Benchmark draws on 99 billion web and app sessions across 6,500 or more websites and nine industries between Q4 2024 and Q4 2025, and it reports conversion rate as a minus 5.1% year-over-year change rather than a level. A publisher sitting on that much data choosing to publish a delta instead of an absolute rate says levels vary too much across the sample to be useful. Read the trend it reports, not a number to measure your own funnel against.

What causes most shopping cart abandonments?

Baymard Institute's survey of US online shoppers puts 40% of abandonments on extra costs such as shipping, tax and fees, 20% on delivery being too slow, and 19% on not trusting the site with credit card information. The page doesn't disclose the survey's sample size. Extra costs are the single biggest reason, and unlike the 70.22% headline rate, this breakdown points at something you can find in a session recording and fix this week.

What is a typical Shopify store conversion rate?

Littledata's 2023 study of 2,800 Shopify sites found an average conversion rate of 1.4%. That figure carries the same limits as any cross-company average. It says nothing about how those stores defined a conversion or what traffic fed them. Treat it as a rough industry reference, not a target for your own store.

How long should my baseline period be for comparing funnel changes?

Use a four-week baseline on your own funnel, with the step definitions frozen before you ship anything. That window controls for the traffic mix, step boundaries and sequencing rules an external benchmark leaves undefined, so any later movement bigger than your normal week-to-week variance is worth investigating. Set the baseline first, then read each step against it after you ship.