14-Step Funnel Audit: Finding Drop-Offs Before They Hurt You

Why We Did This

The client came to us with a general feeling: "Onboarding isn't working, but we don't know where it breaks." They had conversion data at the top level but zero visibility into which specific steps were causing friction. We needed a detailed funnel audit to move from guessing to knowing.

The 14-Step Funnel

The trading app's onboarding flow was actually composed of 14 distinct, measurable steps:

1. Install app — Gets past launch screen
2. First open — Actually opens the app (not just installed)
3. Sign up tap — Clicks the signup button
4. Email entry — Enters a valid email
5. Password creation — Completes password step
6. Mobile OTP sent — Receives and enters OTP
7. Personal details — Name, DOB, address filled
8. PAN entry — Enters PAN number
9. PAN verification — PAN matches NSDL records (this is where 24% dropped)
10. Document upload — Selfie + ID proof uploaded
11. KYC approved — Document verification passes
12. Bank account linking — Links a bank account (18% drop here)
13. Bank verification — Small amount test deposit verified
14. First trade-ready — Account fully activated, can place an order

The Three Biggest Drop-Offs

1. PAN Verification (Step 9) — 24% drop

Users entered their PAN fine, but when the system tried to verify it against NSDL records, a large cohort abandoned. Why? The error messages were cryptic. "Verification failed" without context. Some PAN numbers had slight mismatches (extra spaces, formatting issues). Users didn't know what to do.

2. Bank Account Linking (Step 12) — 18% drop

This is where the psychological friction appears. Linking a bank account feels risky to new users. The UI presented a list of 50+ banks with small logos. Users didn't know which bank to choose or worried about sharing banking details. Many just left.

3. Personal Details (Step 7) — 12% drop

The form asked for 7 fields (name, DOB, address, city, state, pincode, occupation). Mobile keyboard context switching was painful. Many users filled in a few fields then abandoned when the form felt too long.

What We Changed

With the data clear, we prioritized in this order:

PAN Verification: We improved the error messaging. Instead of "Verification failed," we said "PAN not found. Check for extra spaces or try again." We also added a manual verification option if the auto-check failed twice. Result: drop rate fell from 24% to 8%.

Bank Linking: We simplified the bank list. Instead of 50+ options, we showed the top 5 by usage (HDFC, ICICI, Axis, etc.) with larger, clearer logos. We also added copy: "Your bank is secure. We only verify, never store your details." Result: drop rate fell from 18% to 7%.

Personal Details: We split the form into two steps on mobile—basic info first (name, DOB), then address details. We also auto-filled city/state based on pincode. Result: drop rate fell from 12% to 4%.

What Signals Matter

When you're auditing your own funnel, look for:

• Drops larger than 10%: Something is broken. Even small friction compounds at scale.
• Unexpected abandonment: If step X feels "simple" but has high drop, there's hidden friction (poor UX, unclear instructions, trust issues).
• Mobile vs. web divergence: If drop rates differ by 5%+ between platforms, it's usually UX. Mobile keyboard, field sizing, form layout.
• Time spent: If users spend 2x longer on a step, they're struggling. Session recordings reveal what.
• Error rates: High validation errors on a field? Copy or UX is confusing users.

The Complete Picture

By step 14, the client was getting 68% overall completion—not bad, but with these fixes, it climbed to 78% in the next 4 weeks. That's 150 more completed accounts per day at their volume. In a market where user acquisition costs are high, that difference is massive.

The key wasn't any single fix. It was seeing the entire flow as a system, identifying the three biggest leaks, and fixing them with precision. Most teams look at the headline number (68%) and call it acceptable. The ones that audit the funnel and fix the bleeds win.