Verb
Back to blog
Use cases2 min read

Users drop off halfway through onboarding. Here is where, and what fixes it

Find the step where users quit, then remove it, default it, defer it, or do it for them.

Quick answer

Users drop off halfway through onboarding at the first step that needs information they do not have to hand or a decision they do not understand: importing data, connecting another tool, setting permissions. Find that step from your funnel and from what users ask, then remove it, default it, defer it until it is needed, or do it for them with an assistant inside the product.

When users quit onboarding halfway, it is tempting to blame motivation. In a complex product it is usually one step. The count holds steady, then falls off a cliff at the same place for everyone, and that place has a reason you can find.

The step where people leave

The steps that lose people share a pattern. They ask for something the user does not have to hand:

  • Data from somewhere else: a CSV export, a list of team members, last year's figures.
  • Access to another tool: an API key, an admin login, a connection someone else has to approve.
  • A decision they cannot make yet: a permission model, a pipeline structure, a naming scheme.

The user leaves to get it, and the moment passes. Nothing in the product brings them back to that step.

Find it in two places

Your funnel. Track each onboarding step as an event and look at where the count falls most. That gives you the where.

What users ask. Read support email, chat and call notes for the first two weeks of new accounts. The questions cluster around the same step, in the users' own words. That gives you the why.

Four fixes, in order of how much they help

  1. Remove it. Ask whether the step is needed before the user sees any value. Often it is not.
  2. Default it. Pick the setting most customers end up with, and let people change it later.
  3. Defer it. Ask for the data or decision at the moment it is needed, not on day one.
  4. Do it for them. An assistant inside the product can take the request in plain words, such as "add my five teammates as editors", and do the steps through your own features, asking before it changes anything.

Where a tour helps, and where it does not

A product tour helps when users do not know where to click. It does not make a hard step easier, and it does not help someone who left to find a file. For drop-off caused by effort, the fixes above work better. Product tours vs an AI agent goes into the difference.

The verdict

Find the one step where people leave, from your funnel and from what they ask. Then remove it, default it, defer it, or do it for them. Measure the same funnel two weeks later; the change should be visible at that step. Feature adoption rate covers what to measure after onboarding.

Frequently asked questions

Why do users drop off during SaaS onboarding?

Usually at one specific step, not gradually. It is the first step that asks for something they do not have to hand, like a CSV export or an API key from another tool, or a decision they cannot make yet, like a permission model. They leave to find it and do not come back.

How do I find where users drop off?

Two sources. A funnel of each onboarding step shows where the count falls. What users ask, in support email, chat and calls, shows why. The step with the steepest fall and the most questions is the one to fix first.

Will a product tour fix onboarding drop-off?

Only if the problem is not knowing where to click. When users quit because a step is hard or needs their own data, a tour points at the step without making it easier. Removing, defaulting or doing the step works better there.

Vishnu

Vishnu, founder of AskVerb @itsvishnups

Building the AI agent that lives inside a SaaS product and does the work its users ask for. Writing down what I learn along the way, including the parts I got wrong first.

More articles

Give your SaaS its own AI agent

It calls your own API as the signed-in user, confirms before it changes anything, and logs every action. Start with a free 30-day pilot, then we price it on what it actually does for your users.