For founders
Every extra form field costs you users
Baymard Institute's checkout research puts the average US checkout at 14.88 form fields, when 7 to 8 would do, and finds forced account creation behind 18% of abandoned carts. Your sign-up form is a checkout. Here is how to cut it.
Prashant Kumar Singh · · 3 min read
In short
- The average US checkout shows 14.88 form fields; Baymard's research says a checkout can work with 7–8.
- 18% of US shoppers who abandoned a cart in the last quarter did so because the site wanted them to create an account.
- Every field is a question, and each one asks the user to justify the product's curiosity. Ask only for what you need to deliver the first result.
Baymard Institute has spent years testing checkout flows with real shoppers. Two of their findings apply to every product with a sign-up form, not only shops. The average US checkout displays 14.88 form fields, and Baymard's benchmark says a checkout can function with 7 to 8. And in their survey of reasons for abandoning a cart, 18% of shoppers said it was because the site wanted them to create an account.
A sign-up form is a checkout. The user is paying with time and personal data, and the product is asking for the payment before showing the goods. Every field is a reason to leave.
Why fields cost more than they look
A field is not a text box, it is a question, and each question carries three costs. The user has to understand it, has to have the answer to hand, and has to decide whether the product deserves it. "Company size" fails the third test for most people: it is for your sales team, not for them. "Phone number" fails all three. Fields that exist to help the product rather than the user are the ones that get the form abandoned.
The cut, field by field
| Field | Keep? | Why |
|---|---|---|
| Yes | The product needs a way to reach the user; nothing else is required at this stage | |
| Password | Later | A magic link or OAuth button gets the user in without one |
| Name | Later | Ask once they are inside, or infer from email, or do not ask |
| Company, role, team size | No | Sales qualification questions belong after value, not before it |
| Phone | No | Unless the product sends SMS, this is a reason to leave |
| "How did you hear about us" | No | Attribution is your problem, not the user's |
What to do instead of asking
- Show the result before the form. One worked example above the fold tells the user what they are signing up for. Then the email field is a fair trade.
- Let people try before the wall. A single use without an account, then the sign-up, converts better than the wall first. The user has already seen value.
- Infer what you can. Company from the email domain, timezone from the browser, name later from the user's own profile edits.
- Ask the sales questions after the first success. A user who has just got a result will answer three questions. A user staring at a blank form will not.
A note on account creation
Baymard's 18% figure is about forced account creation specifically. The fix is not to remove accounts; it is to make them a consequence of use rather than a precondition. Guest checkout in commerce, and in software a single email field that creates the account silently and lets the user set anything else later. The user never feels the wall because there is none.
Questions people ask
- How many form fields does the average checkout have?
- Baymard Institute's research puts the average US checkout at 14.88 form fields, while its benchmark finds a checkout can function with 7 to 8.
- How much does forced account creation hurt conversion?
- In Baymard's survey of US shoppers who abandoned a cart, 18% named being asked to create an account as the reason.
- What fields should a sign-up form have?
- Email, and little else. Password can be replaced by a magic link or OAuth; name can be asked later; sales qualification questions belong after the user has seen value.
If you're a founder
Send me your sign-up or onboarding flow and the free teardown will come back with the field-by-field cut and what to show before the form.
If you run an AI lab or data team
Task instructions have the same problem: every unnecessary constraint is a field the author has to fill. Specs that ask only for what the test checks get followed more accurately.
Sources