Why this matters
Funnel exploration shows where users drop off across steps like visit → sign up → activate → purchase. As a Data Analyst, you will:
- Identify the biggest conversion blockers and quantify impact
- Prioritize experiments and UX fixes by step-level loss
- Monitor changes after a release using like-for-like funnels
- Communicate trade-offs between top-of-funnel growth and downstream quality
Real-world tasks you might own
- Build a weekly signup funnel and annotate releases that shift conversion
- Segment drop-offs by device, traffic source, or country
- Estimate revenue lift if Step 2 conversion improves by 5 percentage points
Concept explained simply
A funnel is a sequence of user actions. At each step, some people drop off. Funnel exploration quantifies: how many reached each step, what percent converted to the next, and where the losses concentrate.
Mental model
Think of a leaky pipe. Water enters, some leaks at each joint, and the final output depends on cumulative leakage. Fixing the worst leak can increase total flow the most.
Key terms and quick formulas
- Step conversion rate: next_step_users / current_step_users
- Cumulative conversion: final_step_users / first_step_users
- Drop-off count at step i: users_at_step_i - users_at_step_i_plus_1
- Drop-off rate at step i: drop_off_count / users_at_step_i
- Relative improvement needed: target_rate - current_rate
Worked examples
Example 1: Basic conversion and drop-off
Data (week): Visit: 10,000; Sign up: 2,500; Email confirm: 2,000; First purchase: 600.
- Step conversion rates: Visit→Sign up = 2,500/10,000 = 25%; Sign up→Confirm = 2,000/2,500 = 80%; Confirm→Purchase = 600/2,000 = 30%.
- Cumulative conversion: 600/10,000 = 6%.
- Largest absolute drop-offs: Visit→Sign up loses 7,500; Confirm→Purchase loses 1,400.
Why this matters
Top-of-funnel has the biggest absolute loss, but improving the last step may be more revenue-efficient per user. Choose based on feasibility and business goals.
Example 2: Segmenting by device
Mobile vs Desktop at Confirm→Purchase step: Mobile: 300/1,400 = 21.4%; Desktop: 300/600 = 50%.
Insight: Mobile checkout underperforms. Prioritize mobile UX fixes; run AB tests there first.
Example 3: Impact sizing
If Visit→Sign up improves from 25% to 28% with all later rates unchanged:
- Sign ups = 10,000 * 28% = 2,800
- Confirms = 2,800 * 80% = 2,240
- Purchases = 2,240 * 30% = 672
- Lift = 672 - 600 = +72 purchases (+12%).
Mini task: What if Confirm→Purchase rises to 33%?
600 becomes 10,000 * 25% * 80% * 33% = 660. Lift: +60 (+10%).
Step-by-step method
- Define the funnel steps clearly (event names, order, time window).
- Choose a consistent cohort (e.g., users who started in the same week).
- Count unique users per step; avoid double-counting retries.
- Compute step and cumulative conversion; visualize as bars.
- Segment by meaningful dimensions (device, channel, country).
- Identify the highest-impact step (biggest loss or most valuable lift).
- Propose experiments and estimate potential impact.
- Monitor post-change using the same definitions.
Consistency checklist
- One user counted once per step
- Same time window for all steps
- Same event definitions across weeks
- Drop-offs only counted between adjacent steps
Hands-on: Exercises
These mirror the Exercises below. Work through them here, then submit your answers in the exercise block if you want to compare with the solution.
Step Users Landing page views 8,000 Account created 2,400 Email confirmed 1,920 Onboarding completed 1,344 First purchase 672
- Compute each step conversion rate and the cumulative conversion to purchase.
- Identify the step with the highest drop-off rate and the largest absolute drop-off count.
- If Email confirmed→Onboarding completed rises by 5 percentage points, how many additional purchases result (assume all later rates unchanged)?
Show a structured approach
- Write rates as decimals to chain easily (e.g., 0.3 not 30%).
- Use both absolute losses and rates to choose priorities.
- When simulating lift, change only one step; keep others constant.
Before you check the solution
- Did you compute both step and cumulative conversion?
- Did you separate absolute vs percent drop-offs?
- Did you hold other steps constant when simulating the improvement?
Common mistakes and self-check
- Mixing cohorts: Comparing users who started in different weeks. Self-check: Are all counts from the same start cohort?
- Counting events, not users: One user can trigger multiple events. Self-check: Are you using unique users per step?
- Changing definitions mid-analysis: Inconsistent events across weeks. Self-check: Does your SQL or tool use the same filters each time?
- Ignoring sample size: Tiny segments show noisy rates. Self-check: Flag segments with low counts and avoid strong claims.
- Optimizing the wrong step: Improving a small-loss step yields little gain. Self-check: Size the impact before proposing fixes.
Who this is for and prerequisites
Who this is for: Aspiring and practicing Data Analysts, Product Analysts, Growth Analysts, and PMs needing clear funnel insights.
Prerequisites: Comfort with basic percentages and ratios; familiarity with user events; optional SQL basics or spreadsheet skills.
Practical projects
- Weekly signup-to-purchase funnel with device segmentation and a one-page summary of insights
- Channel-level funnel comparing paid vs organic and a recommendation of where to optimize first
- Post-release funnel comparison: before vs after, holding definitions constant, with a lift estimate
Learning path
- Start: Funnel basics (this page)
- Next: Cohort analysis and retention curves
- Then: Experimentation basics (A/B test readiness and guardrail metrics)
- Finally: Automated dashboards and alerting for funnel health
Next steps
- Complete the exercise and check the solution
- Take the quick test below to confirm understanding
- Build a simple funnel dashboard for a product you know
Note: The quick test is available to everyone; if you are logged in, your progress will be saved.
Mini challenge
A funnel has 20,000 Visits → 3,000 Signups → 1,800 Activations → 540 Purchases. You can improve only one step by 4 percentage points. Which step likely maximizes purchases and why? State your choice and a short rationale.