Redesigned the signup form and tripled conversion, delivered with AI in 5 hours
edrone is a marketing automation CRM for ecommerce. Automated messages and newsletters bring shoppers to the store to buy: a cart they left behind, a first order, the next one.
For ten years edrone sold one way only, through Sales. Freemium was the company's first product-led channel, and this work landed four months into it.
I diagnosed the drop from event data and session recordings, then bet the redesign on one judgment: that splitting the form and creating the account on the first step would raise conversion despite adding a step. It did. I designed the flow and built its frontend in Codex.
The four-step funnel lost the most people between clicking "Sign up free" and creating an account. A mandatory phone number almost nobody used stayed in the form for one salesperson who cold-called quiet signups, and every signing-up user paid for it at the most expensive moment in the funnel.
"Sign up with Google" did not create an account. It pulled a name and email from the Google dialog and dropped the user back on the same four fields, now partly filled. The button looked like a shortcut and behaved like autofill.
I started in Amplitude, checked that the event data was sound, then watched session recordings of that exact step to see what people were doing on the form. That gave me a dozen or so issues I thought were wrong.
Alongside it, I ran a UX agent I'd built on Claude through the flow. It walks the screens in a browser, clicks through them the way a user would, and comes back with a report: what's wrong, why, and how severe.
I used the agent to test my own hypotheses rather than to produce them. It confirmed most of what I had already found, added candidates I had not considered, and ranked them. I picked the ones worth the time and cost of building.
Time to act on what the research had found.
The main change was removing the phone number input. I had to confirm with Sales whether it held real value; it turned out to be marginal, worth cutting outright.
Next, SSO finally worked correctly, so users could create an account through Google or Shopify.
And that's where the third change came in: creating an account still required the shop's URL up front...
So I split that step into two, an unpopular move, though it did streamline the SSO flow. After creating the account, the user supplied their shop link and name, for the Support team and in-app personalization.
I put together a prototype draft in Figma, and after checking with a developer it turned out the whole thing needed some backend work, so I took the front end myself, in Codex.
And that's how we shipped the new signup flow in five hours.
I waited three weeks before deciding whether the flow stayed or not. Given the site's average traffic, that was the least I'd trust the number, though honestly, I already knew the decision after the first few days of seeing the performance.
Desktop, about 85% of the traffic, went from 1% to 2%; mobile, which had a dismal 0.05%, rose to 3%.
- Desktop
- Mobile
| Channel | Before | After |
|---|---|---|
| Desktop | 1% | 2% |
| Mobile | 0.05% | 3% |
At the same time, I was aware of the risk that rolling out three changes at once could make it hard to tell which one actually moved the needle, especially splitting the form into two steps, so I tracked that one on its own. Share of visitors who began filling the form went from 2.7% to 10%: desktop 3% to 10%, mobile 1% to 10%. Splitting the form was the bet, and this is what confirms it.
For cleanliness, I would have shipped the phone number removal on its own, then SSO and the split together as a second phase, but I didn't have that much time at that point.