Setting Up Analytics: Events, Funnels, and Cohorts

From Gut Feel to Ground Truth

Knowing which metrics matter is only useful if your product actually captures the data behind them. Before a PM can look at a dashboard and say "Day 7 retention dropped," someone had to decide what to track, name it consistently, and wire it into the product. That is the job of setting up analytics.

Most modern products use a product analytics tool layered on top of (or instead of) traditional web analytics. The three you will meet most often are:

A bank like Access Bank or a fintech like Moniepoint might use GA4 for their marketing website and Mixpanel or Amplitude inside the actual banking app, where every tap, transfer, and bill payment can be tracked as an event.

The Building Block: Events

Everything in product analytics starts with an event — a single, named action a user takes. Good event tracking follows a simple grammar: Object + Action.

• Signup Completed
• Transfer Initiated
• Transfer Successful
• Card Added
• Airtime Purchased

Each event can carry properties — extra details attached to it:

Event: Transfer Successful
Properties: {
  amount: 15000,
  currency: "NGN",
  recipient_bank: "GTBank",
  transfer_type: "own_account",
  time_to_complete_seconds: 4.2
}

Properties are what let you slice a metric later: "Show me transfer success rate for transfers over ₦100,000" or "Compare completion time between iOS and Android."

The Tracking Plan

Before an engineer writes a single line of tracking code, a PM should write a tracking plan — a spreadsheet listing every event, its properties, and when it fires. This prevents the common mess of five engineers naming the same action TransferDone, transfer_complete, and Payment Success in three different places, which makes the data useless six months later. Consistency is more valuable than cleverness.

Funnels: Where Do Users Drop Off?

A funnel is a sequence of events a user is expected to complete in order, with the percentage who make it from one step to the next.

Example — a Paystack-style checkout funnel:

1. Checkout Page Viewed — 100%
2. Payment Method Selected — 78%
3. Card Details Entered — 61%
4. OTP Submitted — 54%
5. Payment Successful — 49%

Reading this funnel, a PM immediately sees the biggest leak is between step 2 and step 3 (17 percentage points lost). That is where to focus research: is the card form confusing? Too many required fields? Slow to load on 3G networks common outside major Nigerian cities?

Funnels turn a vague complaint ("checkout feels slow") into a specific, measurable, prioritized problem.

Cohorts: Grouping Users by What They Have in Common

A cohort is a group of users who share a defining characteristic, most often the week or month they signed up. Cohort analysis lets you compare groups fairly, since a user who joined yesterday hasn't had time to churn yet.

Beyond signup-date cohorts, PMs also build behavioral cohorts:

• Users who made their first transfer within 24 hours vs. after a week
• Users acquired via referral vs. paid ads
• Users on Android vs. iOS

Comparing retention curves across these cohorts often reveals which acquisition channel or onboarding path produces the healthiest long-term users — insight a single blended number can never show. Netflix famously uses cohort analysis to see how content releases affect the retention of subscribers who joined in different months.

Putting It Together

A mature analytics setup gives a PM the ability to ask a question and get an answer in minutes rather than filing a ticket and waiting a week for an engineer to query the database. That speed of learning compounds — teams that can answer "did that change work?" quickly ship better products faster.