A funnel is an ordered sequence of steps a user is expected to complete on the way to some goal — view product, add to cart, start checkout, enter payment, complete purchase. At each step, some fraction of users fail to continue to the next one. Funnel analysis is the practice of measuring that step-by-step drop-off rate so you know exactly where, in the sequence, the largest fraction of potential value is being lost.

The fundamental output of a funnel report isn't the overall conversion rate — it's the conversion rate *between each pair of adjacent steps*. A 2% overall conversion rate tells you almost nothing actionable; a funnel showing 80% → 60% → 55% → 20% → 18% tells you exactly where to focus.

Fix the worst step, not the first step

Teams instinctively focus on the top of the funnel because it has the most raw traffic. But the highest-leverage fix is usually the step with the steepest percentage drop, regardless of its position. A step that loses 65% of users who reach it is a bigger opportunity than doubling top-of-funnel traffic into an already-leaky funnel.

Open vs. closed funnels

A closed funnel requires users to complete steps in strict order — step 2 only counts if it happened after step 1 for that user. An open funnel counts anyone who completed each step, in any order or combination, and simply reports the aggregate step counts. Closed funnels are more accurate for diagnosing a single sequential flow (like checkout), but they undercount if your product allows legitimate step-skipping (e.g., a returning user with a saved cart skipping 'add to cart'). Open funnels are more forgiving but can overstate conversion by not enforcing sequence or timing.

GA4's funnel exploration defaults to open funnels but gives you a toggle for 'make open funnel' — leaving it unchecked makes it closed and ordered.

Open vs closed funnel behavior

Funnel type	Step completion rule	Best for
Closed, strict order	Step N only counts if step N-1 happened first, in sequence, for the same user	Linear flows like checkout where order is enforced by the UI
Open, any order	Each step counted independently per user, regardless of order or timing	Exploratory analysis where strict sequence isn't guaranteed

Diagnosing why a step drops, not just where

A funnel report tells you *where* the drop-off is; it never tells you *why* on its own. Once you've identified the worst step, you diagnose the cause with other tools: segment the funnel by device, browser, or traffic source to see if the drop is universal or concentrated (a 70% mobile-only drop at the payment step usually means a broken mobile form); check session replay or heatmaps for that specific step to see users physically struggling; and check page load time and error-event logs for that step, since a slow or erroring page will show up as a funnel drop long before anyone complains.

Segmenting the drop-off step is the single highest-value next action after spotting a funnel problem — an undifferentiated drop rate almost always hides a segment-specific one.

funnel_dropoff.txt

text

Step                 Users    Step conv.   Cumulative conv.
1. view_item         50,000     100%           100%
2. add_to_cart        22,000     44%            44%
3. begin_checkout     19,800     90%            40%
4. add_payment_info    7,900     40%   <-- worst step    16%
5. purchase            7,100     90%            14%

Largest single-step loss: step 3 -> 4 (payment info), not step 1 -> 2
despite step 1->2 having a lower percentage, because the ABSOLUTE
user count lost at step 3->4 combined with its severity (60% loss on
a late, high-intent step) represents more recoverable revenue.

Funnel exploration in GA4

GA4's Funnel Exploration report lets you define a custom sequence of event-based steps, visualize drop-off as a horizontal bar chart per step, and break down any step by a dimension (device category, source/medium, new vs. returning) directly inline. It also supports 'elapsed time' between steps, which surfaces a different failure mode — users who complete every step but take an abnormally long time at one of them, suggesting friction even without an outright drop.

Unlike the standard reports, funnel explorations aren't retroactive by default for custom steps defined after the fact on events that weren't tracked with the needed parameters — the step definitions rely on events and parameters that must already exist in your data to be usable.

Funnels assume your events are complete

A funnel built on a missing or unreliable event will show a fake drop-off that's actually a tracking gap, not a user behavior problem. Before trusting a funnel report, verify each step's event fires reliably across browsers and devices — a funnel step that silently fails to track on Safari will look identical to a step users are actually abandoning.

Trend over time, not just a single snapshot

A single funnel snapshot is useful, but the real value compounds when you track a specific step's conversion rate over time after making a change — redesigning a payment form, removing a required field, adding a progress indicator. Treat each funnel fix the way you'd treat an A/B test: define the step's baseline conversion rate first, ship the change, and measure whether the step's conversion rate — not just overall revenue — actually moved.

What's next

A funnel assumes one straight-line path, but real users bounce across channels and sessions before converting. Multi-channel funnel reporting addresses how credit for that conversion should be distributed.

Next: Multi-Channel Funnel Reporting →

I build these systems professionally.

Whether it's a RAG pipeline, analytics migration, or AI workflow — let's talk.

Work With Me →

Need production AI or MarTech expertise? I build custom LLM workflows, server-side GTM tagging, and analytics infrastructures professionally.

Need custom AI or MarTech setup? Let's build together.

Let's talk →