Funnel Analysis for Signup Flows and Why Users Drop Off

To find why users drop off in your signup funnel, define each step as a tracked event, measure conversion between steps by device and source, then watch the sessions of people who left at the worst step and ask a few of them directly. Funnel analysis shows where. Replays and questions show why.

Signup funnels leak in predictable places: the first form field, email verification, a payment step, a permission request, an onboarding screen that asks for too much too early. The hard part is not knowing that leaks exist. It is finding the specific one costing you the most and understanding it well enough to fix it. Here is a method that works whether you run a SaaS trial, a marketplace or an online store with accounts.

Step 1. Define the funnel from the user's point of view

Write the steps as the user experiences them, not as your database records them. A typical B2B SaaS signup might be:

1. Visits pricing or a landing page
2. Clicks "Get started"
3. Submits email and password
4. Enters the verification code
5. Completes the onboarding questions
6. Reaches the first moment of value (first project created, first report seen)

Two rules help. Keep every step a single, unambiguous event, and stop the funnel at a moment of real value rather than at "account created". An account nobody uses is not a conversion.

Step 2. Measure conversion between steps, then segment

Look at step to step conversion, not only the overall rate. A funnel that converts 8 percent end to end could have one terrible step or six mediocre ones, and the fixes are completely different.

Then split every step by:

• Device and browser. Mobile problems are the most common hidden cause of drop off. A code input that does not trigger the numeric keyboard, or a modal taller than the screen, can quietly halve a step.
• Traffic source. Paid campaigns often bring visitors with different expectations. If one channel drops at the pricing step, the ad may be promising something the page does not deliver.
• Country and language. Phone verification, address formats and payment methods fail differently by region.
• New versus returning. Returning visitors who still do not sign up are telling you something specific.

The worst step for the biggest segment is your starting point. A funnel analysis tool should let you build this view in minutes and keep it saved for the next sprint.

Step 3. Watch the sessions of people who left

Numbers tell you where to look. Recordings tell you what happened. Filter session replays to visitors who reached the problem step and did not reach the next one, then watch ten to twenty of them. Take notes in plain sentences.

Things to look for:

• Fields that are filled, cleared and filled again (unclear format or validation)
• Rage clicks or dead clicks on the submit button (disabled state with no explanation)
• Switching tabs right after the email step (they went to find the code and did not come back)
• Long pauses on a single field (a question they did not want to answer)
• Error messages that appear below the visible area on mobile

Form analytics adds field level detail: time per field, refill rate and the last field touched before abandonment. If one field is the last interaction for a large share of leavers, you have found your suspect.

Step 4. Ask the people who left

Behavior narrows the options. A direct question confirms which one is true. Show a short, one question survey to visitors who abandon at the problem step, or invite recently lost signups to a brief interview. Keep it specific. "What stopped you from finishing signup today?" works better than a generic satisfaction score.

In UserExplorer you can target a survey or an AI interview at exactly the users who showed the behavior, and their answers land next to their sessions in the same profile. See how it works on the AI user research and surveys and NPS pages.

Step 5. Fix one thing and measure again

Ship the smallest change that addresses the cause you found, then compare the same step for the same segment before and after. Give it enough traffic to be meaningful. If the step improves and the next step does not get worse, keep the change and move to the next leak.

A worked example from start to fix

Here is how the five steps fit together on a typical trial signup. The numbers are illustrative, not data from any real product.

A team sees that end to end conversion from "Get started" to first project is lower on mobile than on desktop. The step view shows the gap sits almost entirely between "submits email" and "enters verification code". Segmenting further, the drop is worst for visitors arriving from a paid social campaign.

Watching twenty sessions of mobile visitors who stopped at that step shows a clear pattern. They submit the form, switch to their mail app, and many never return. Some of those who do return find a fresh page asking them to start again, because the verification screen was not kept in the session. A one question survey shown on the next visit confirms it: the most common answer is that the code email took too long or could not be found.

The fix is small. The team keeps the verification state when the visitor returns, adds a clear resend link, and supports one tap code fill on mobile. The same step for the same segment is compared over the following weeks. It improves, and the next step does not get worse, so the change stays and the team moves to the onboarding questions.

Mistakes that make funnel data lie

• Counting internal traffic. Your own team, test accounts and automated checks pass through the signup flow more often than you think. Exclude them, or every step looks healthier than it is.
• Changing event definitions mid period. If "account created" moved from the form submit to the verified email last month, the trend line shows a drop that never happened.
• Using too short a window. Some people verify their email the next morning. A funnel that only counts completion within thirty minutes will call them lost.
• Ignoring repeat attempts. A visitor who fails, leaves and returns two days later can appear as both a drop and a conversion. Decide whether you measure visitors or attempts, and be consistent.
• Reading tiny segments. A step that falls from ten conversions to six in one country is not a trend. Wait for enough traffic before you act.

Common causes of signup drop off and their usual fixes

Doing this faster with an AI researcher

The method above works with any set of tools. It is also slow, because steps two to four mean jumping between a funnel report, a replay tool and a survey tool, exporting segments and watching recordings one by one.

UserExplorer was built to shorten that loop. You can ask the AI researcher a question such as "why do mobile trial users stop at verification" and get a plain English answer with the evidence sessions attached, so you can check the reasoning yourself before you change anything. Every finding links back to real recordings, never to a guess.

To see what such a report looks like, enter your site address on the homepage and open the sample report. Plans for teams of every size are on the pricing page.