Funnel Analysis: Transition Rates, Drop-Off Rates, and GA4 Steps
Published:
Last Updated:
Category: Marketing Glossary, Marketing Strategy
Authors: Shusaku Yosa
Funnel analysis follows a consistently defined population through ordered steps toward an outcome. A transition rate is the number progressing to the next step divided by the number at the preceding step; drop-off rate is one minus that rate. These calculations require matching populations, units, order, and time conditions. Independent monthly totals do not automatically form a valid funnel.
Fix the unit and success conditions first
Do not mix users, sessions, and event counts between steps. Dividing cart-addition events by product-viewing users includes repeat actions in the numerator and is not a user transition rate. Document the KPI definition and aggregation conditions before building the chart.
Design item | Specify | Example check |
|---|---|---|
Population and unit | Users, sessions, or another consistent unit | Use the same deduplication rules |
Step condition | Event and required parameters | Product view, add to cart, successful purchase |
Order and time | Sequence and allowed completion window | Purchase within seven days of viewing |
Entry rule | Start at the first step or permit later entry | Closed versus open funnel |
Success | A completed outcome rather than a click alone | Order accepted or inquiry saved successfully |
A submit-button click can precede validation or network failure. Define conversion success conditions and check failed sends, reloads, resubmissions, and duplicate events.
Calculate a closed funnel
A closed funnel follows users entering at the first step and progressing in the specified sequence. This fictional example uses one population and counting unit, without later-step entrants.
Step | Users | Transition from previous step | Users not progressing |
|---|---|---|---|
Entry | 10,000 | Not applicable | Not applicable |
Product view | 4,000 | 40% | 6,000 |
Add to cart | 800 | 20% | 3,200 |
Checkout | 400 | 50% | 400 |
Purchase success | 200 | 50% | 200 |
Product-view-to-cart transition is 800 / 4,000 = 20%, with 80% drop-off. Overall success is 200 / 10,000 = 2%. A zero denominator makes a transition rate undefined, rather than automatically zero.
Drop-off means the next defined step was not reached under the rules. It does not necessarily mean the user left the website. They may continue researching, switch devices, finish outside the allowed window, or have missing measurement.
Separate later entry in an open funnel
An open funnel allows users to enter at later steps. Consequently, all downstream users divided by all upstream users may not be a transition rate. In GA4, users enter at their first qualifying step and subsequent progress follows the specified sequence. Skipping a required step prevents later steps from counting in that sequence. See the official funnel exploration guide.
Consider four fictional users: A→B→C, B→C, A→C, and C only. The A→C user skips B, so C is not counted as continued completion from A.
Step | Closed-funnel users | Open-funnel users |
|---|---|---|
A | 2 | 2 |
B | 1 | 2 |
C | 1 | 3 |
Open-funnel C / B is 3 / 2 = 150%, but that is not the B-to-C transition rate: it includes one user entering directly at C. Both users reaching B continued to C, giving 100% for that population. Keep entrants separate when calculating progression.
Configure a GA4 funnel exploration
- Create a funnel exploration and set dates and population.
- Define steps with event or dimension conditions, rather than confusing metrics with step rules.
- Select closed entry for a first-step cohort or open entry to investigate later entrants.
- Choose whether the next step must directly follow or may have intervening events.
- Set a step-completion time limit appropriate to the purchase or decision cycle.
- Add a breakdown and save the rules with the resulting counts.
Direct-following conditions can be affected by other collected events between steps. Check the actual sequence rather than making the funnel unintentionally restrictive. GA4 also uses the first qualifying sequence for a user within the period; purchase event totals need not equal final-step user counts.
Breakdowns use the user’s first qualifying value. A user entering on mobile and finishing on desktop is not necessarily reassigned to a new device category at each step. The GA4 dimension and metric definitions help clarify counting units.
Prioritize beyond the largest drop-off percentage
In the fictional closed funnel, the 80% product-to-cart drop-off is prominent, but the entry-to-product step loses more people: 6,000 versus 3,200. Consider reach, achievable change, downstream completion, value, effort, and measurement confidence. A KPI tree makes the assumptions connecting a step to the final outcome explicit.
If cart transition hypothetically rises from 20% to 25%, with 4,000 product viewers and both later rates unchanged at 50%, the calculation is 4,000 × 25% × 50% × 50% = 250 purchases. That is 50 above 200, but it is a conditional fictional scenario, not a forecast or promised uplift.
Document validation and missing coverage
Record the issue, causal hypothesis, change, primary outcome, and guardrails. Before-and-after comparisons can be affected by seasonality and acquisition mix. Where feasible, randomly assign a defined experimental unit to comparison groups, and set duration, metrics, and stopping conditions in advance.
B2B sales and offline purchases also require stage-handoff definitions and time-lag treatment. For OMO journeys connecting online and stores, join only data that can legitimately and reliably be linked. Cross-device and offline actions are not automatically one person’s complete journey. A missing funnel step is not sufficient evidence of staff failure or poor content.
