Funnel Analysis in Mobile Analytics — What It Is, Funnel Stages and How to Build

What Is Funnel Analysis?

Funnel Analysis is an analytics method in which a sequence of events leading to a conversion is identified. Each subsequent step includes only those users who completed all previous ones. The result is a “funnel” — a wide entrance (everyone who started) and a narrow exit (those who completed).

In mobile apps, a typical funnel looks like this: install → registration → onboarding → first action → repeat action → purchase. At each step, some users drop off, and the analyst’s task is to understand the cause and reduce the losses.

Basic Funnel Metrics

The Average Time metric between steps is often underestimated. If it takes an average of 45 minutes from install to registration, the problem lies in content loading or the complexity of the first screen.

Funnel Types in Mobile Analytics

Funnels are divided into two main types — open and closed. The choice depends on the scenario and the flexibility of the user journey.

Strict Funnel

The user must complete all steps strictly in order. If they skip a step or take another path, they are not included in the funnel. Example: registration → email confirmation → profile fill → first order. Strict funnels provide a clean picture but exclude real user scenarios.

Open Funnel

The user can perform steps in any order or skip some. The system counts conversion at each step independently. Example: catalog view → add to cart → checkout. Open Funnel is a more realistic method for most mobile apps.

• Strict funnel — strict sequence, all steps mandatory
• Open funnel — free order, steps optional
• Hybrid — first steps mandatory, subsequent ones free

Hybrid Funnel Example for Subscription Apps

Step 1 (mandatory): app install. Step 2 (mandatory): subscription screen view. Step 3 (optional): tariff plan view. Step 4 (mandatory): subscription purchase. The hybrid approach balances data purity and realism.

Stages of Building a Funnel

Building a funnel is a task at the intersection of analytics and product. The process consists of five steps, from hypothesis formulation to report automation.

Step 1: Define the Target Conversion

What do you want the user to do? purchase_completed, subscription_started, referral_sent. The target event is the top of the funnel. All steps below lead to it.

Step 2: Break Down the Path into Events

List all mandatory events between app entry and the target action. For e-commerce: app_opened → product_viewed → add_to_cart → checkout_started → payment_selected → purchase_completed. Each event should already exist in Event Tracking or be added.

Step 3: Choose a Time Window

Time window — the maximum time between the first and last step. For quick scenarios (game install) — 24 hours. For subscription-based — 7 or 30 days. Events outside the window are not included in the analysis.

Window choice directly affects conversion: too short a window (1 hour) cuts off users who returned after 2 hours to complete a purchase. Too long (30 days) includes random actions. The optimum is determined by analyzing the time between steps: build an interval distribution and select the window at the 90th percentile level.

For fast funnels (onboarding, registration) use a 1-hour window. For medium (first purchase) — 7 days. For long (subscription) — 30 days. Different windows for different step pairs can be used within the same funnel.

-- Example of a Funnel calculation in BigQuery
WITH funnel_steps AS (
  SELECT
    user_id,
    event_name,
    event_timestamp,
    LAG(event_name) OVER (
      PARTITION BY user_id ORDER BY event_timestamp
    ) AS previous_event
  FROM `project.events.*`
  WHERE event_name IN (
    'app_opened', 'product_viewed',
    'add_to_cart', 'purchase_completed'
  )
)
SELECT
  event_name AS step,
  COUNT(DISTINCT user_id) AS users,
  ROUND(
    COUNT(DISTINCT user_id)
    / MAX(COUNT(DISTINCT user_id)) OVER (), 3
  ) AS conversion_rate
FROM funnel_steps
GROUP BY step
ORDER BY ARRAY_POSITION(
  ['app_opened', 'product_viewed',
   'add_to_cart', 'purchase_completed'],
  event_name
)

The query uses LAG to track the sequence and ARRAY_POSITION to sort the steps. The result is a table with user counts and conversion at each step.

Step 4: Validate on Historical Data

Build the funnel on data from the last 30 days. If “checkout_started” has 80% conversion but “purchase_completed” has 25%, the problem is in the payment form. Compare with competitive benchmarks: average cart conversion in mobile apps is 15–25% according to Statista (2024).

Step 5: Automate Monitoring

Set up a daily report with the funnel in a dashboard (Looker Studio, Tableau). Add alerts: if the Conversion Rate at a step drops below the control limit, send a notification to Slack or Telegram.

Drop-off Analysis in the Funnel

Drop-off Analysis is a key stage of Funnel Analysis. Losses at each step are a source of hypotheses for product improvement. Three problem zones are identified: technical bugs, UX barriers, and motivational scenarios.

Technical Losses

If 15% of users drop off at the “registration_completed” step after entering their phone number, check the SMS gateway. Technical losses can be fixed in 1–2 days and provide a quick conversion boost. Use Firebase Crashlytics in combination with the funnel to identify bugs.

UX Barriers

If 40% is lost at “cart_viewed” → “checkout_started”, the problem is in the cart UX. Perhaps there is no “Checkout” button above the fold, or the delivery cost is not shown before entering the address. UX analysis requires session replays (Hotjar, UXCam, Smartlook).

Motivational Losses

If the user reaches “payment_selected” but does not complete the purchase, the value of the offer is lower than the price. Motivational losses are addressed with discounts, trials, and social proof. A/B test: show 50% of users reviews on the payment page and compare conversion.

• Technical — bugs, timeouts, API errors (up to 20% of losses)
• UX barriers — complex navigation, hidden CTAs (up to 35%)
• Motivational — price, lack of trust (up to 45%)

Analysis Methodology: 5 Whys for the Funnel

The 5 Whys technique applied to the funnel: at a step where conversion dropped by 20%, sequentially ask five “why” questions. Why don’t users reach payment? Because the payment form loads slowly. Why does it load slowly? Because the request to the payment gateway times out. Why does it time out? Because the SDK is not updated. The methodology identifies the root cause, not the symptom.

Cohort analysis of the funnel is another tool. Compare the funnel of users who installed the app in January and February. If the February cohort shows a 15% drop in conversion at the “registration” step, the cause is a change in the onboarding screen rolled out in late January. Cohorts isolate temporal factors.

According to CXL Institute (2024), systematic Drop-off Analysis followed by A/B testing of hypotheses yields an average funnel conversion increase of 30% over 3 months.

Tools for Funnel Analysis

Each tool offers its own approach to building funnels. The choice depends on product complexity and budget.

Amplitude — Pathfinder and Funnel Analysis

Amplitude is the leader in product analytics. Its built-in Funnel Analysis allows segmenting the funnel by cohorts, devices, and app versions. Pathfinder shows branching — how many users took an alternative path after step 2.

Mixpanel — Flows

Mixpanel Flows visualizes the funnel as a transition graph. Pro: real-time updates. Con: limited filtering in the free tier. Suitable for mid-sized projects with a team of up to 10 analysts.

Firebase Analytics — Funnel Report

Firebase provides a free Funnel Report with a limit of up to 10 steps. Integration with BigQuery allows building custom funnels via SQL. Ideal for startups just beginning to analyze conversion.

Tableau + BigQuery

For enterprise solutions, use Tableau with BigQuery as the source. This provides full flexibility: custom metrics, end-to-end analytics from install to LTV, real-time monitoring. Requires a BI engineer.

Google Analytics 4 — Funnel Exploration

GA4 provides Funnel Exploration in the Explore section. Supports up to 10 steps, segmentation by user parameters, and time period comparison. Con: limited flexibility compared to Amplitude, but completely free and integrated with Google Ads.

Funnel Analysis Tools Comparison

Tool choice is determined by analysis frequency. For daily funnel monitoring, Amplitude or Mixpanel works. For weekly strategic reports — Tableau with BigQuery. For startups — Firebase or GA4.