Form Field A/B Tests: How Many Fields, Which Ones

Ask for as few form fields as you need to qualify and follow up on the lead.

For most lead-gen and SaaS sign-up forms, that means three or four.

In our database, cutting fields that weren't needed beat the longer form again and again: Kareo got +30% new doctor sign-ups, and InsightWhale got +26% conversions by keeping three required fields.

But a field that makes the form easier to finish, or that filters out bad leads, can win too.

This hub covers form tests on SaaS and lead-gen sites: sign-up, demo, quote, application and enquiry forms.

Every matching test shows in the grid on this page, 61 of them as of November 2026. What follows is the analysis: what won, what lost, and how to run your own test. For checkout forms, see checkout flow tests.

Reducing form fields: what happened when teams cut them

Removing a field is the most common form test, and the most dependable one we have.

Here are the clearest results from the database.

The Qualicorp result is the cleanest of these.

Only one thing changed, the test ran 14 days on a 50/50 split, and the fields removed were asking for data visitors had already typed.

If your form asks for anything twice, fix that before you test anything else.

InsightWhale's test is worth copying for its design.

It had three versions: the original eight required fields, three required fields with the optional ones shown below, and three required fields with the optional ones hidden under a dropdown.

The hidden version won. So people were reacting to how long the form looked, not only to what they had to type.

Read the smaller wins carefully. Kommunicate measured button clicks, not finished sign-ups, so some of that +25.53% may have dropped off on the next screen.

Mercantil Bank changed the copy as well as the fields, and the result was published by a vendor with no sample size.

Bundled redesigns point the same way, though they can't isolate the fields.

BusinesSuites cut a six-field quote form to three and stripped the page around it for +88% conversions.

The team also noted that leads from the new page behaved differently later in the pipeline.

Watch for that whenever you make a form easier.

Form fields and conversion rate: when more fields won

"Fewer fields always wins" is not what the data shows.

Several tests added a field, or made one required, and won.

Qualicorp's profession field had been optional, with a "none" choice that led to empty quote results.

Making it required, with more options to choose from, raised completed requests by +12.52% over seven weeks.

The extra effort helped because without that field the quote didn't work.

An older MarketingExperiments test added an email field to a free-trial call to action and told visitors why it was there.

Trial sign-ups went from 0.68% to 1.57%, +133%, across about 63,000 unique visitors.

Kommunicate's test above went the other way on the same idea.

The difference may be that this page gave a reason for the field.

It's also possible that sites and eras just differ.

Removing a choice can also hurt. When a subscription funnel dropped its term-length dropdown and locked everyone into monthly billing, conversion fell about 40%.

A field that lets people pick what suits them is not friction.

Some fields did nothing either way. Patrick McKenzie removed two mailing-list checkboxes from a sign-up form and saw no change in trials.

That's a useful flat result: if a field doesn't hurt, keep whichever version serves the business.

The clearest loss for adding a field came from MarketingSherpa and Act-On.

A landing page offered a free guide for four required fields.

Adding a topic choice alone made no difference.

Adding the choice plus one more required field caused a significant drop against both other versions.

The extra field, not the choice, did the damage.

Which fields matter: hard fields cost more than extra fields

Field count is a rough measure. Some of the biggest wins came from fixing one field that was hard to fill in, while the count stayed the same.

• Information people don't have to hand. Tinkoff Bank added a "Don't remember passport details?" option so applicants could send them later. Applications rose +35.8% across more than 265,000 visitors.
• Awkward date inputs. Vandebron replaced a date-of-birth calendar that defaulted to 2000 and moved one month at a time. The new field showed the format and added slashes as people typed. Sign-ups rose +16.3%, and drop-offs on that field fell 80%.
• Long dropdowns. PureGym swapped a dropdown of 450+ gyms for a search field and got +3.4% conversion. Casa Mineira went the other way: it replaced a free-text property search with two dropdowns for +57.25% leads on about 7,500 visitors. Short lists suit dropdowns; long lists suit search.

Before you remove fields, look at your form analytics for the one where people stall or quit.

Fixing that field is often a bigger win than cutting two easy ones.

Form design: labels, markers and layout

Small changes to how fields look have won, mostly by making it clearer what the form needs.

• Micro Focus added asterisks to required fields and a "required field" note on a free-trial form: +5.89% completions over 20 days.
• Springest replaced asterisks with "optional" labels and added social proof in the same test: +5.21% on over 2,500 conversions. Because the changes were bundled, the labels' share of the lift is unknown.
• Qualicorp's quote form got +16.93% completed requests with floating labels.
• Teamleader added icons next to its sign-up fields: 32.84% to 37.11%, or +13.1%.

The Micro Focus and Springest results fit Baymard Institute's checkout research, which recommends marking both required and optional fields.

Baymard found that marking only one type led to validation errors and confusion.

Clever layouts did worse. Patrick McKenzie tested a "madlibs" form, with the fields placed inside a sentence, against a standard labeled form.

The madlibs version lost: 21.73% vs 27.55%, about 22% fewer trial sign-ups, at 95% confidence.

The test was short (34 hours, 736 people), but it's a good warning not to copy a trend without testing it.

Form length and conversion: steps, placement and the whole page

How long a form feels depends on how many fields people see at once and where the form sits on the page, not only on the total field count.

Splitting a long form into steps

Reassured turned a long single-page life insurance quote form into one question per step with a progress bar.

Submissions rose +31.23% over about five weeks. Apexure split a claim form into four stages with the easy questions first, and conversion went from 1.77% to 4.33%.

EDP moved required fields later in its funnel, closer to the offer, for more than 25% more leads.

More steps can lose too. RegOnline went from a two-step to a three-step trial sign-up that asked for contact details first, and completed leads fell 24.5%.

Britannica cut a whole page from registration and got +117% free trials and +130% revenue.

Steps help when each one is easy and people can see they're making progress.

Steps that are only there to collect more data hurt.

Our multi-step form tests cover this in depth.

Baymard's checkout research makes the same point from a different angle.

The average US checkout in its 2024 benchmark had 11.3 form fields, while most sites need only 8. Baymard also found that the number of fields affects usability far more than the number of steps.

Where the form sits

Placement results go both ways. Tom's Planner put a four-field sign-up form above the fold on its homepage for +44% sign-ups.

Vendio did the opposite: it moved the form off its landing page to a separate step and got +60% sign-ups.

Northmill hid a loan form behind an Apply button for +7% applications.

A form on the page works when visitors arrive ready to act.

A button that leads to the form works better when they still need persuading, or when the form takes up so much space that it crowds out the reasons to sign up.

Haller Enterprises shows why you need the right metric.

It replaced a five-field hero form with booking and call buttons.

Form leads dropped 72%, but booking value per visitor rose +24.49%.

If the team had measured only form submissions, they would have picked the wrong version.

What lost, and the bias in this evidence

Of the 61 matching tests, 55 are recorded as wins, 3 as losses and 3 as flat.

That ratio says more about what companies choose to publish than about form tests in general.

Losing and flat tests rarely get written up.

So treat each result as proof that a change can work, not as the odds that it will work for you.

The losses share a pattern: they added effort without giving the visitor anything back.

That covers the extra required field at MarketingSherpa, the added step at RegOnline, the madlibs layout, and the removed term choice.

Many wins were also bundled with copy or layout changes, and some were before/after comparisons rather than split tests.

The Venture Harbour multi-step form result (0.96% to 8.1%) came from years of iteration, and the company later sold a form tool.

How to run your own form field test

1. List every field and who uses it. If no one in sales or onboarding acts on a field, it's a candidate for removal or for moving after sign-up.
2. Find the problem field. Use form analytics to see where people pause, make errors or leave. Start with fields that are repeated, hard to fill in, or ask for information people don't have to hand.
3. Test one change at a time when you can. If you remove three fields, you won't know which one mattered. Removing them all at once is fine if you only need a quick answer.
4. Measure past the submit button. Track qualified leads, demos held, trials activated or revenue as well as submissions. Fewer fields can bring in more leads that are worse, as BusinesSuites noticed.
5. Size the test before you start. Form tests often have small effects. Take a sign-up form converting at 27.55%. To detect a 10% relative lift (to about 30.3%) at 95% confidence and 80% power, you need roughly 4,250 visitors per variant. At a 5% baseline, the same lift needs about 31,000 per variant. Run your own numbers in the sample size calculator and check results with the significance calculator.

A sample-size calculation is the wrong tool if you're making a huge change, such as cutting a form from twelve fields to three.

A big effect will show up quickly. In that case, the bigger risk is lead quality, not statistics.

For more lead-gen form ideas, see our guides to lead form optimization and SaaS sign-up flows, or browse all tests and other page elements.