Guide

How to diagnose low SaaS conversion

Low conversion is a pile of different failures wearing one label. Diagnose it in order. Skip a step and you will spend the month on the wrong fix.

1. Use one window and one definition

Last 30 days, or one cohort. Same source for every count. Visitors from the analytics tool you actually trust, signups from the database, activated users from the event you would bet the product on, new customers from billing. Mixing “all time signups” with “this month’s traffic” produces a rate that cannot be low or high. It is just inconsistent.

2. Compute the four rates before you have an opinion

Visitor → signup, signup → activation, activation → paid, visitor → paid. Write them down. If 500 people visit and nobody signs up, another feature probably isn’t the first experiment to run. The empty step is the top of the funnel. If 40% of visitors sign up and almost none activate, the landing page is not the argument.

3. Throw out steps with no sample

A rate needs a base. Under a few dozen people, the percentage will twitch every day. Mark that step “not judged” and look at the step above it. Buying ads to “get a real sample” is reasonable only when the earlier steps are not already obviously broken.

4. Name one cause, then a result that would kill it

“Improve conversion” is not an experiment. “New users who see a finished example invoice will activate more often, and if they don’t after the next 50 signups, onboarding length is not the cause” is an experiment. One change. One metric on the weak step. A failure signal you will believe.

5. Do not run the other fixes yet

Price, positioning, and a new channel can all be wrong at once. You still learn nothing if you change them together. Park the ones that sit on a step you have not shown is thin.