Redesigned the signup form and tripled conversion, delivered with AI in 5 hours
Problem
Freemium was edrone's first product-led channel, and four months in, acquisition was the focus. But most people who clicked "Sign up free" left the form without creating an account. Signup conversion sat at 0.75%, far below the 2 to 5% typical for the industry.
edrone is a marketing automation platform for online stores. It sends automated emails and newsletters that bring shoppers back to buy, for example after they leave a full cart or to nudge them towards their next order.
Research
To find out why people left, I went through the flow three ways:
- Analyzed the signup flow's event data in
Amplitude - Watched session recordings of the form
- Ran my self-built UX Audit Agent on
Codex through the flow
Amplitude
CodexSolution
The research pointed to three changes that could lift conversion:
- Removed the phone number input, after Sales confirmed it was worth little.
- Made Google and Shopify sign-in create the account, instead of only filling in the form.
- Split the form into two steps, with the account first and the shop's name and URL after. It adds a step, but it lets SSO work without asking for the URL first.
I built the frontend in Codex and shipped the new signup flow in 5 hours.
Outcome
After 3 weeks of live traffic, signup conversion had tripled. The biggest jump was on mobile, which went from almost no signups to converting better than desktop.
| Channel | Before | After |
|---|---|---|
| Desktop | 1% | 2% |
| Mobile | 0.05% | 3% |
Shipping three changes in one release made it hard to tell which one moved the number. With more time I would have shipped them separately.
To hedge, I added extra tracking in Amplitude. It clearly confirmed the split, as the share of visitors who began filling the form went from 2.7% to 10%.