SaaS Signup Funnel: Find Where Users Drop Off
Every funnel leaks. The question is where it leaks most. Map each step from first visit to paying customer, measure the share that moves on, and fix the biggest drop before touching anything else.
How do you find where users drop off?
List every step from first visit to paying customer, track how many people reach each one, and calculate the share that moves to the next. The step with the biggest drop relative to what’s normal for you is the one to fix first. Then watch real users at that step to see why they stop.
What does a typical funnel look like?
A product-led or trial funnel usually has six steps. The example uses fictional numbers to show how a funnel table reveals the biggest leak; in this case, most signups never reach the first meaningful result.
| Step | People | Moved on |
|---|---|---|
| Visited the site | 10,000 | – |
| Started signup | 600 | 6% |
| Completed signup | 420 | 70% |
| Reached first result (activated) | 105 | 25% |
| Active in week four | 60 | 57% |
| Became paying customers | 24 | 40% |
How do you fix each kind of drop-off?
Each step has typical causes and fixes. Before changing anything, watch session recordings or sit with a few users at the problem step; the reason is often obvious once you see it.
| Drop-off | Common causes | Fixes to try |
|---|---|---|
| Visit to signup start | Unclear value, wrong visitors, weak call to action | Message match, clearer headline, better targeting |
| Signup start to complete | Long form, verification friction, slow or shifting page | Fewer fields, social login, fix speed and layout |
| Signup to first result | Empty product, too many setup steps | Templates, sample data, guided first task |
| First result to habit | Value isn’t repeated, no reminders | Usage-based emails, team invites |
| Active to paying | Price or plan mismatch, no clear upgrade trigger | Upgrade prompts at value moments, sales assist |
Do technical issues affect the funnel?
Yes, especially at signup. Pages that load slowly or move while loading lose people. Google treats a Cumulative Layout Shift below 0.1 and an Interaction to Next Paint under 200 milliseconds as good (Google’s Core Web Vitals documentation). Test signup on a phone, on mobile data, at least once a month.
Fix the leak, then pour in more.
More traffic into a leaky funnel raises cost, not customers. Fix the biggest drop-off before increasing spend.
How should you measure it over time?
Define each step the same way every month, track it by acquisition source, and review it on a fixed schedule. A source that brings many signups but few activations is costing more than it looks. For funnel stage definitions in a sales-led model, see MQL, SQL and pipeline stages. If visitors don’t start signup at all, start with traffic but no signups; if the audience is wrong, see why GTM isn’t working. Lead magnets can add a softer entry point: lead magnets for B2B SaaS. Deeper conversion work is in the SaaS Conversions guide, starting with free trial activation.
Frequently asked questions
What is a SaaS signup funnel?
The steps from first visit to paying customer: visit, signup start, signup complete, activation, habit and payment.
How do I find where users drop off?
Track how many people reach each step, calculate the share that moves on, and watch users at the step with the biggest drop.
What is activation in SaaS?
The first moment a user gets real value from the product, such as completing the core task.
What is a good signup conversion rate?
It varies too much by definition and traffic source to compare across companies. Improve on your own baseline.
Should I fix traffic or conversion first?
Conversion. More traffic into a leaky funnel raises cost without raising customers.
Does page speed affect signups?
Yes. Slow or shifting pages lose people, especially on mobile.
How often should I review my funnel?
Monthly, by acquisition source, using the same step definitions each time.
Sources & further reading
Figures and rules change. These sources let you verify the current information directly. Example numbers on this page are illustrative, not benchmarks.