Funnel Step Definition

What is Funnel Step Definition?

Funnel step definition is the precise, unambiguous specification of each step in a funnel: the event to match, who is counted, in what order, within what time limits, and with which filters. Clear definitions make your funnel metrics trustworthy and reproducible.

Why this matters

• Product decisions: Identify which step blocks users (e.g., Add to Cart vs. Checkout Started).
• Experiment analysis: Ensure A/B results aren’t biased by loose or changing definitions.
• Stakeholder alignment: Avoid debates caused by different interpretations of “conversion.”
• Diagnostics: Pinpoint where drop-offs happen and which cohorts are most affected.

Concept explained simply + Mental model

Think of a funnel like a series of doors. A user must pass each door in order, within a time limit, and with the right badge (filters). A step definition tells the guard at each door exactly who may pass and under what conditions.

• Stream: A user generates a stream of timestamped events.
• Gates: Each step is a gate with rules (event name, properties, filters).
• Clock: A time window controls how long a user has to reach the next step.
• Counter: Decide whether you count unique users, sessions, or orders.

Core components of a step definition

• Event: The specific event name(s) to match.
• Actor identity: user_id, session_id, account_id (state clearly).
• Order: Steps are usually sequential; clarify if out-of-order is allowed or not.
• Time window: Maximum allowed time between steps (e.g., 24 hours after previous step).
• Filters: Event/property filters (platform, country, AB variant, device), user segment filters, exclusion rules (e.g., staff/bot traffic).
• Counting unit: Unique users vs. sessions vs. orders. State one.
• Deduping rule: Use first event after the previous step, or nearest, or last within window.
• Entry rule: Where users can enter the funnel (first step only, or multiple entry points).
• Optional/OR steps: Allow optional steps or alternate paths? Specify clearly.
• Resetting: Does a user re-enter the funnel after completing it (or after timeout)?

Worked examples

-- Pseudocode for clarity only
WITH s1 AS (
  SELECT user_id, MIN(timestamp) AS t1
  FROM events
  WHERE event='signup_started' AND platform IN ('web','ios','android') AND staff=false
  GROUP BY user_id
), s2 AS (
  SELECT e.user_id, MIN(e.timestamp) AS t2
  FROM events e JOIN s1 ON e.user_id=s1.user_id
  WHERE e.event='email_verified' AND e.timestamp BETWEEN s1.t1 AND s1.t1 + INTERVAL '7 days'
  GROUP BY e.user_id
), s3 AS (
  SELECT e.user_id, MIN(e.timestamp) AS t3
  FROM events e JOIN s2 ON e.user_id=s2.user_id
  WHERE e.event='first_session' AND e.timestamp BETWEEN s2.t2 AND s2.t2 + INTERVAL '3 days'
  GROUP BY e.user_id
)
SELECT COUNT(DISTINCT user_id) AS activated FROM s3;

Decision rules and edge cases

Exercises

Do these short exercises. Then open the solutions to self-check.

{
  "count_by": "user_id",
  "entry": "step1_only",
  "reentry": "no_within_7d",
  "steps": [
    {"id":1, "event":"app_install", "filters":{"country":["US","CA"],"bots":false}},
    {"id":2, "event":"app_open", "window":"7d_after_prev", "dedupe":"first"},
    {"id":3, "event":"tutorial_completed", "window":"24h_after_prev", "dedupe":"first"},
    {"id":4, "event":"account_created", "window":"24h_after_prev", "dedupe":"first"}
  ],
  "optional_between": [{"after":3, "before":4, "event":"enable_notifications"}]
}

Common mistakes and how to self-check

• Missing time windows: Leads to inflated conversions. Always state a window between steps.
• Ambiguous actor identity: Mixing session_id and user_id changes rates. Pick one per funnel.
• Ignoring duplicates: Repeats can jump steps or skew attribution. Define deduping.
• Vague entry rule: Allowing mid-funnel entry without saying so produces inconsistent cohorts.
• Changing definitions mid-report: Document version and freeze it during analyses.
• Not excluding test/staff/bots: Pollutes data. Add explicit exclusions.

Practical projects

• Project 1: Redefine an existing funnel with precise windows, deduping, and exclusions. Compare conversion rates before vs. after.
• Project 2: Build an OR-branch funnel (e.g., Activation via Invite OR Project Creation). Show how branch logic changes step-through rates.
• Project 3: User vs. Session funnel. Implement both on the same path and explain differences in drop-offs.

Mini challenge

Define a subscription funnel: View Pricing → Start Checkout → Submit Payment → Subscription Active. Constraints: out-of-order events are possible due to delayed webhooks; payment must be within 48h of checkout; subscription_active may arrive up to 24h after payment; count by account_id; exclude currency != USD; allow re-entry only after a cancellation. Write your exact rules.

Who this is for

• Product Analysts and Data Analysts working on growth, activation, and revenue
• PMs or Marketers who need consistent conversion metrics
• Engineers defining tracking specs

Prerequisites

• Basic event analytics concepts (events, properties, users/sessions)
• Comfort reading simple SQL or analytics tool filters
• Understanding of your product’s key events and data quality caveats

Learning path

• 1) Learn event schema and naming conventions
• 2) Practice defining steps with windows and filters
• 3) Handle duplicates, optional/OR steps, and re-entry
• 4) Validate with timelines and sample users
• 5) Automate and document funnel definitions

Next steps

• Apply these rules to one live funnel this week
• Document your final definition and freeze it for experiments
• Set a recurring review to catch drift in events or properties

Quick Test

Take the quick test below to check your understanding. Everyone can take it for free; only logged-in users will have their progress saved.