Why Users Abandon Forms (and Why It’s Rarely a Design Problem)
A pet cremation app dropped hundreds of users at step two. The fix wasn’t a redesign. It was answering twelve questions nobody wanted to ask out loud.
A woman opened the app at 2am to arrange her dog’s cremation.
She got to step two of four. Then she stopped.
I know this because I was looking at the analytics the next morning, and she was one of hundreds doing the same thing at the same step. Our dashboard had a name for it: friction. Red bar, big number, fix it.
The standard playbook here is obvious. Fire a recovery email. “You left something in your cart.” Maybe a discount.
Except the thing in her cart was her dog.
Form abandonment usually starts with an unanswered question
We had built this platform the way we build most things. A marketplace. Clinics on one side, pet owners on the other, a clean four-step form in the middle. It worked, in the sense that people used it.
But when we finally sat down and asked what she was actually trying to find out at that exact field, the answer wasn’t about the form at all.
Is it individual or communal? Do I have to bring him myself? Can I keep the collar?
She didn’t abandon the form. She hit a question and there was nobody to ask. It was 2am. She wasn’t going to phone anyone. She probably wasn’t going to phone anyone at 10am either, because these are the questions people find embarrassing to say out loud.
That reframe changed everything we did next.
Why we stripped out the progress bar, the upsell and every nudge
We had spent months building for a user in a hurry. Someone who wants to get through the flow fast and get on with their day.
This user wasn’t in a hurry. She was in pain, and she was careful.
So we cut everything that assumed urgency. The progress bar came out. The upsell came out. Every reminder nudge, every “complete your booking” push, gone.
I remember the discomfort in that meeting. We were removing things that had taken real work to build. Nobody likes deleting features.
The twelve-question chatbot that beat every redesign
The least glamorous item on the roadmap. A chatbot.
Get vikash soni’s stories in your inbox
Join Medium for free to get updates from this writer.
Not an AI concierge. Not “conversational commerce” or whatever the pitch deck word was that year. Twelve questions, answered clearly, available at 2am.
We built the list by reading support tickets, listening to clinic staff, and looking at where people paused. Twelve questions people wanted answered before they could take the next step, and would rather not ask a human.
That feature outperformed every redesign we shipped on that product. Every single one. The platform runs across 650+ pet aftercare clinics today, and the thing that moved the number most was not a new interface. It was answering questions.
The mistake: building a marketplace before understanding the user
We built it as a marketplace first and asked about the user’s emotional state about six months too late.
I don’t say that to be dramatic. Six months of engineering went into optimising a flow for the wrong person. If we had spent one week early on asking “what does she feel at step two,” we’d have skipped most of it.
Chat converts 5x better than forms, and the reason is not the chat
Here is the pattern I now see in almost every product I touch.
Website forms convert at roughly 2 to 3 percent. Chat conversations convert at 10 to 20 percent. Same visitor, same intent. The difference is whether someone answered the question standing between them and the next step.
Every funnel has a step where users go quiet. Analytics calls it friction, and sometimes it is. A slow page, a confusing label, too many fields. Fix those.
But often it’s not friction. It’s an unanswered question. And no amount of UX polish fixes a question. You can make the button bigger and the form shorter and the animation smoother, and she will still stop at step two because she doesn’t know if she can keep the collar.
A one-afternoon test to find hidden questions in your funnel
This takes an afternoon. I’ve run it on fintech onboarding, on a healthcare booking flow, on a B2B quote form. It works every time.
- Pull your highest drop-off step.
- Write down the three things a user wants to know at that exact moment. Not what you want to tell them. What they want to know.
- Check whether your product answers those three things on that screen.
If it doesn’t, you don’t have a design problem. You have a silence problem.
This leaks the most revenue in high-consideration categories. Health, legal, insurance, anything expensive or emotional. These are the categories with the most questions and the least willingness to ask them out loud. Which means the gap between your form and your chat conversion is probably bigger than 2 percent versus 10.
Start with your worst drop-off step this week
Open your analytics. Find your worst step.
Then ask the only question that matters: is this friction, or is it something you never answered?
I’ve written up how we built the twelve-question list, including the four we got wrong the first time. If you want it, leave a comment and I’ll send it over.
