User Versus Session Funnels

Why this matters

As a Product Analyst, you routinely diagnose drop-offs and conversion changes. Choosing the wrong unit of analysis (user vs session) can reverse conclusions, mislead prioritization, and waste engineering time.

• Weekly growth review: diagnose a signup dip. Is it fewer users converting or more low-intent sessions?
• Experiment readout: an onboarding change affects first-session behavior. Should the funnel be session-based?
• Retention deep-dive: are returning users progressing over multiple sessions or stuck per session?

Concept explained simply

User-based funnels count unique people. Each person can appear at most once per step. Use when your question is about people progressing over any time window.

Session-based funnels count attempts. A person can have multiple sessions and multiple attempts. Use when your question is about within-session behavior or experiences that reset each visit.

Mental model

Think of a supermarket:

• User funnel = how many distinct shoppers eventually buy after entering the store this week (even if they come twice).
• Session funnel = how many store visits result in a purchase (each visit is an attempt, regardless of shopper identity).

Key trade-off: User funnels show reach and long-term progression; session funnels show friction within a single attempt.

Definitions & when to use each

• User funnel
  • Denominator: unique users who hit Step 1 (in the analysis window).
  • Each user counted once per step, even across multiple sessions.
  • Best for: onboarding completion across days, long journeys (discovery → purchase), activation/retention.
• Session funnel
  • Denominator: total sessions that hit Step 1.
  • Each new session is a new attempt.
  • Best for: first-session onboarding, checkout flow within a visit, support flow per incident, kiosk or POS usage.

Worked examples

Example 1: Signup funnel

Window: last 7 days

• Unique users who visited landing: 800
• Unique users who started signup: 400
• Unique users who completed signup: 320
• Total sessions with landing: 1,200
• Sessions with start signup: 500
• Sessions with complete signup: 360

User-based conversion

• Start rate: 400 / 800 = 50%
• Complete (overall): 320 / 800 = 40%
• Step-to-step: 320 / 400 = 80%

Session-based conversion

• Start rate: 500 / 1,200 ≈ 41.7%
• Complete (overall): 360 / 1,200 = 30%
• Step-to-step: 360 / 500 = 72%

Interpretation: Sessions underperform users. Many people complete eventually, but not necessarily in their first attempt. Focus on first-visit friction.

Example 2: Product → Cart → Purchase

Window: last 14 days

• Users who viewed product: 600 (users)
• Users who added to cart: 300
• Users who purchased: 240
• Sessions with product view: 900 (sessions)
• Sessions with add to cart: 320
• Sessions with purchase: 260

User-based: Add rate 300/600 = 50%; Purchase overall 240/600 = 40%; Step-to-step 240/300 = 80%

Session-based: Add rate 320/900 ≈ 35.6%; Purchase overall 260/900 ≈ 28.9%; Step-to-step 260/320 = 81.3%

Interpretation: Strong eventual conversion per person, but many low-intent views inflate sessions. Consider traffic quality and pre-qualification.

Example 3: In-app support flow (incident-based)

Events: Crash → Open Help → Submit Report

• Users with any crash: 1,000
• Users who submitted a report: 500
• Sessions with crash: 1,600
• Sessions with report: 720

User-based: 500/1,000 = 50% of affected users eventually report.

Session-based: 720/1,600 = 45% of crash sessions get a report in that session.

Interpretation: If the goal is to improve response per incident, session funnel is correct. If you want the proportion of users who ever report, use user funnel.

How to build funnels correctly

1. Choose the entity: user or session. Write it at the top of your query or notebook to avoid confusion.
2. Define the window: Within-session funnels often use minutes; user funnels can use days/weeks between steps. Decide max gap between steps.
3. Counting rules
   • Repeated steps: do you take first, last, or best attempt?
   • Out-of-order events: require strict order or allow repeats in between?
   • Cross-device identity: ensure user deduplication logic is sound.
4. Attribution: If multiple sessions touch the funnel, decide where to attribute completion (first session, last session, or any).
5. Quality filters: Exclude bots, test users, and clearly incomplete sessions (e.g., < 3s duration).

Common mistakes (and self-check)

• Mismatched denominators: Step 1 uses users, Step 2 uses sessions. Self-check: print count type next to each step.
• Infinite windows: Users complete after months; you think the funnel is great. Self-check: cap max days between steps and report both capped and uncapped.
• Double counting repeats (session funnel): One session fires Add to Cart twice. Self-check: deduplicate by session_id and step.
• Identity drift: Cross-device users counted as multiple people. Self-check: compare results using stable identifiers (e.g., account_id).
• Attributing to wrong attempt: Purchase attributed to first session instead of last. Self-check: list sample paths and verify attribution rule.

Exercises

Do these before the quick test. A short dataset is included for manual calculation.

u1, s1, visit, 00:01
u1, s1, start_signup, 00:02
u1, s1, close, 00:03
u1, s2, visit, 10:00
u1, s2, start_signup, 10:01
u1, s2, complete_signup, 10:03
u2, s3, visit, 02:00
u2, s3, start_signup, 02:02
u3, s4, visit, 05:00
u3, s4, start_signup, 05:01
u3, s4, close, 05:03
u3, s5, visit, 07:00
u4, s6, visit, 08:00
u4, s6, start_signup, 08:02
u4, s6, complete_signup, 08:05

Mini challenge

Marketing launched a campaign that increases top-of-funnel traffic. Conversion rate dropped on a session-based funnel, but user-based conversion stayed flat. In one paragraph, explain what likely happened and one action you would take next.

Practical projects

• Build a dual-entity signup dashboard: toggle between user and session funnels with the same steps; include overall and step-to-step rates.
• Run an A/B test readout template: show first-session conversion (primary) and user conversion within 7 days (guardrail).
• Create a data quality validator: checks entity consistency, dedup rules, and window caps for any funnel query.

Who this is for

• Product Analysts and Data Analysts diagnosing conversion and onboarding.
• PMs who need to interpret funnel metrics correctly.
• Growth marketers evaluating campaign impact on conversion.

Prerequisites

• Basic event data concepts: user_id, session_id, timestamp.
• Comfort with percentages, ratios, and time windows.
• Ability to group and deduplicate records.

Learning path

1. Review event tracking and identity basics.
2. Learn funnel step definitions and ordering.
3. Master user vs session funnels (this lesson).
4. Practice with mixed funnels (hybrid windows, attribution variations).
5. Apply to experiments and retention studies.

Next steps

• Complete the exercises below and check your answers.
• Take the quick test to validate understanding. Test is available to everyone; only logged-in users will have progress saved.
• Apply this to a real product flow you know; produce both user- and session-based results and compare insights.

Progress note

Your quick test progress is saved only if you are logged in. You can still take the test for free.