All articles Fraud & ATO

SaaS Signup Funnel Analysis, Step by Step

Engineer studying a translucent blue funnel of stacked layers on a dark monitor in a black studio

A SaaS signup funnel analysis answers a narrow question your traffic dashboard usually dodges: at which step does a visitor stop becoming a user, and how long do they hesitate before they do. Sessions tell you what came in. The funnel tells you where the path broke.

The method is mostly discipline. Pick the event that means “signed up,” draw at most eight steps from landing page to that event, and read two numbers per step: conversion and drop-off. Then add the number most teams skip, the median time between steps.

One condition up front: the sessions you count must exclude bots, or every percentage downstream is wrong.

Define the signup event before you draw the funnel

Name one event and freeze it

Most funnel arguments are definition arguments. Write down the single event that means “signed up” — usually the server-side event that creates the account row, not the button click — and use it in every report. If the client fires signup_submit and the server fires account_created, you have two funnels with two different answers.

Pick the starting step per question

A funnel starting at “any page view” answers how many visitors ever sign up; one starting at the pricing page answers how well pricing converts. Both are valid and not comparable. Say which one you are reading.

Count quality, not only quantity

Add a second funnel ending at “email verified” or “first project created.” The gap between the raw signup and the qualified one is where typos, trial abuse and automated form fills live.

Choose two to eight steps that match real decisions

Every step should be a moment where the user decides or does something: viewing pricing, starting the form, submitting it, verifying the email, creating the first object. Prynt Analytics runs funnels of two to eight steps, roughly the number of decisions a normal signup flow contains.

Hands arranging six glowing translucent tiles in a row on a black glass surface

Too many steps hide the drop; if a funnel has fifteen, the one that hurts is buried between two that matter less. A workable sequence: landing page view → pricing view → signup form start → signup submit → email verified → first project created.

Two rules. Do not split one screen into two steps unless a real event sits between them, and do not count micro-interactions such as tooltip opens — they inflate the top of the funnel and dilute the drop.

Read conversion and drop-off at every step

Conversion for a step is the users who completed it divided by the users who reached it; drop-off is the remainder. Read them with absolute counts: a 40% drop from twelve users is noise, a 15% drop from four thousand is a project.

Separate “didn’t finish” from “couldn’t finish”

Both look like drop-off and need different fixes. Didn’t finish is intent — the visitor compared prices and left. Couldn’t finish is friction — a validation error with no message, an OTP that never arrived, a script that threw on the last field.

Replay tells them apart fastest. Prynt Replay masks every form field in the browser and again on the server, with no unmask option, so you can watch the session and still see where it broke.

Measure the median time between steps

Median, not average: one user who left the tab open overnight drags an average far from what most people experienced.

Person in silhouette watching a dark display of descending blue bars in an analytics room

  • Long time, little drop-off. Usually fine. Email verification is the classic case: people confirm later and come back.
  • Short time, heavy drop-off. Suspect confusion or blocking — a disabled button, a silent rejection, a redirect loop.
  • Two clusters. Half finish in seconds, half take minutes: returning users who autofill, and newcomers reading every field.

Set a threshold. Any step whose median time is well above the others is the first one you inspect, before you touch the top of the funnel.

Exclude bot sessions before you trust a number

Signup pages attract automation: form fillers, credential stuffing, crawlers, AI agents that submit forms. If they count as sessions, the denominator at step one grows while the numerator does not, and conversion looks worse than the experience your users had. Prynt Analytics excludes bot sessions from every number it reports, so the funnel reflects people.

When a step drops, Identify supplies the signals behind it: visitorId, datacenter and VPN/proxy signals, browser tampering, automation detection. Our bot detection guide covers the signal set, and device fingerprinting explains how events stay linked across sessions.

Turn the worst drop into a policy, not a redesign

Not every drop is a design bug. If your signup step absorbs multi-accounting, trial resets or SMS pumping, the drop may be doing its job, and the useful question is whether the policy is calibrated.

Abstract node network with thin blue lines and dissolving particles against a black background

Prynt Control answers use-case decisions (signup, free_trial, login, checkout, sms, paywall) with allow, step-up or block, plus a reason code and a suggested step such as captcha or verify_email. It starts with a 14-day mirror period in which nothing is blocked, so you compare what the policy would have decided against what converted. Control is a separate plan, not part of the lower tiers.

FAQ

How many steps should a signup funnel have?

Between two and eight, each tied to a decision rather than a page view. Prynt Analytics supports funnels of two to eight steps, which covers most landing-page-to-signup flows without burying the step that hurts.

Should bot sessions count in signup funnel analysis?

No. Automated sessions inflate the top of the funnel and rarely complete it, so leaving them in makes conversion look lower than it is. Prynt Analytics excludes them from every reported number.

Why the median time between steps and not the average?

One outlier drags an average. The median shows what a typical user experienced, so a slow step stands out instead of hiding in a long tail.

Build your first funnel this week

Prynt Analytics is $39/month with a 14-day free trial: overview, live view, custom events, funnels, goals, weekly retention and paths, with bots excluded from every number. Start with three steps, from landing page to signup, then extend the funnel once you trust the definition. Begin at pryntid.com or read the Analytics product page.

Try it free

Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.

Keep reading