Form Optimization: Fewer Fields, Better Answers
Every form field is a small cost for the visitor. How to audit your forms, which fields to cut, and how to find out why people stop halfway.
A form is the place where interest turns into an action: a sign-up, a request, a purchase. It is also where a surprising number of visitors give up. They arrived, they read, they decided to go ahead, and then a form asked for more than they were willing to give.
The good news is that forms are among the easiest parts of a funnel to improve. The changes are small, the effect is easy to measure and you do not need new traffic to see it. This article walks through how to audit a form, what to cut, what to fix and how to find out why people stop.
Every field has a cost
Each field asks the visitor to stop, think and type. On a phone it also means a keyboard, a small target and a chance to make a mistake. A field is only worth that cost if the answer is needed right now.
Baymard Institute studies checkout forms in detail. In its 2024 review the average checkout asked for about 11.3 form fields, while its analysis suggests that most sites need only around eight. Baymard also reports that the number of fields affects the usability of a checkout much more than the number of steps does. The data comes from e-commerce checkouts, but the principle carries over: the question is not how many steps you split a form into, but how much you ask for in total.
Run a field audit
Take the form and list every field in a table. For each, answer three questions:
| Field | Why do we need it now? | Can we get it later? |
|---|---|---|
| Company size | Routing to the right sales contact | Yes, at the first call |
| Phone number | Not needed for a trial | Yes, optional at onboarding |
| Account and confirmation | No |
Be strict with the second column. “It would be nice to know” is not a reason. “We need it to deliver what the visitor asked for” is. Fields that fail the test can be removed, made optional or moved to a later step, for example to onboarding after the account exists.
This idea is sometimes called progressive profiling: ask for the minimum first and collect the rest once the visitor has seen value. It works well for B2B sign-ups, where sales would like to know everything and the visitor would like to try the product.
Fix the details that cause errors
Once the field list is lean, look at how the form behaves.
Labels above fields, always visible. Placeholder text disappears while the visitor types, and then they cannot check what was asked. Keep a real label.
Clear, specific error messages. “Invalid input” helps nobody. “The email address needs an @ sign” does. Show the message next to the field and keep what the visitor already typed.
Validate at the right time. Checking a field while it is still being typed produces errors that are not errors yet. Validate when the visitor leaves the field, and again on submit.
Use the right input types on mobile. An email field should open the email keyboard, a phone field the number pad. Turn on autofill hints so the browser can fill name, address and email.
Do not make people repeat themselves. Ask for the email once, not twice. Offer “same as billing address”. Accept spaces and dashes in phone numbers instead of rejecting them.
One clear button. The label should say what happens: “Request access” tells more than “Submit”. Give the button a visible state while it is sending, and do not let a double click send the form twice.
Think about trust, not only usability
Some fields are not hard to fill in, they are hard to give. A phone number, a company revenue range or a date of birth makes visitors ask why you want it. If you need such a field, say so right next to it in one short line: “We only call if you ask us to.” For anything touching personal data, link the privacy policy near the button and keep the wording plain. Building trust online goes deeper on the signals that make people comfortable.
Measure where the form loses people
You cannot fix what you cannot see. Track three numbers for each form:
- Visitors who see the form.
- Visitors who start it, meaning they focus the first field.
- Visitors who submit it successfully.
The gap between one and two tells you the offer or the page leading to the form is the problem. The gap between two and three is the form itself. If you can, also record which field was the last one touched before someone left. That points at the exact place where it hurts. These start and submit events are classic micro-conversions, and they are available long before you have enough submissions for a clean test.
Ask why people stop
Numbers show where. Only the visitor can tell you why. When someone has started a form and moves toward the exit, a single short question can reveal the reason:
- “What stopped you from finishing?”
- Options such as “Too many questions”, “I wanted to check the price first”, “I am not sure what happens next”, “Other”.
Keep the question short and show it once. The answers often name the fix directly. If many say they do not know what happens after submitting, you need a line under the button, not a shorter form. The article on asking without annoying users covers how to ask in a way that does not become the next obstacle, and Exit Intent: When Leaving Visitors Are Worth Asking explains when the moment is right.
Test one change at a time
When you change a form, change one thing at a time where possible: remove one field, rewrite the button, move the privacy line. If you change everything at once you will not know what worked. Decide the success measure beforehand, usually the share of visitors who start the form and finish it, and let the test run long enough to be meaningful. What to Do Before Your First A/B Test covers the groundwork, and the book by Kohavi, Tang and Xu is the standard reference on running controlled experiments properly.
If traffic is low, a before-and-after comparison combined with visitor feedback is a reasonable substitute. Just write down the date of the change and what else happened that week.
A short checklist
- Does every field have a reason that serves the visitor right now?
- Are optional fields really needed, or can they go?
- Is the label always visible?
- Do error messages say what to fix?
- Does the mobile keyboard match the field?
- Does the button say what happens next?
- Do you know the share of visitors who start and the share who finish?
- Have you asked the people who left?
Work through the list for your most important form this week. Most teams find at least one field that can go and one message that can be clearer, and those two changes alone often move the numbers.
Frequently asked questions
How many fields should a form have?
As few as the task needs. Baymard's checkout research found that most sites need only about eight fields, while the average checkout asks for more than eleven. For a lead form, start with the minimum you need to follow up.
Do multi-step forms convert better than single-page forms?
Not automatically. Baymard found that the number of fields affects usability more than the number of steps. Splitting a long form into steps can help, but it does not replace removing fields.
Should I mark optional or required fields?
Make the common case quiet. If most fields are required, mark the few optional ones. Better still, remove optional fields unless you can name what you do with the answer.
How do I find out why visitors abandon a form?
Measure the start and the submission of the form to see the drop, then ask. A single question shown when a visitor leaves a half-filled form, such as 'What stopped you?', often names the problem faster than a test.
Sources and further reading
- Checkout Flow: Average Form Fields , Baymard Institute
- Trustworthy Online Controlled Experiments: A Practical Guide to A/B Testing , Kohavi, Tang & Xu, Cambridge University Press, 2020