How to increase your form conversion rate

What is form conversion rate, and how do you calculate it?

Real submissions divided by the people who saw the form, times 100. Twelve messages from 400 people who saw the form is 3%.

Both halves of that fraction are easy to get wrong.

The bottom half: people who saw the form, not visitors to the page. If the form sits below three screens of text, most page visitors never reach it, and your rate looks terrible for a reason that has nothing to do with the form. Count a view when the form itself is on screen. Formgong's analytics script does it this way: a view is counted once, when at least 40% of the form is visible, and a start when the visitor first clicks into a field they can fill in. Hidden fields and the spam trap don't count.

The top half: real submissions, not all submissions. We counted every submission to forms on Formgong accounts that aren't ours over the 30 days to 6 November 2026: 154 in total on 30 forms. 28 of them, 18%, were bots caught by the hidden honeypot field, and all 28 went to just 3 of the 30 forms. On the worst of those three, 20 of 24 submissions were bots. If your number jumped and the inbox didn't, check the spam folder before you celebrate a redesign.

The same mix-up explains the “average conversion rate” numbers you'll find online, which run from a couple of percent to nearly half. One article divides by everyone who opened the page, another by people who started typing, a third by visitors to a landing page that is nothing but the form. Those are three different numbers. Don't aim for any of them; compare your form with itself, before and after a change.

First, fix what fails without an error

The most expensive form problems don't look like problems. The visitor clicks Send, nothing visible happens or a thank-you appears, and no message reaches you. While writing this week's articles on form details, we hit five of these in Chromium:

None of them would show up in a heatmap or an A/B test. Fill in your own form on a phone and a computer, send it, and check that the message arrives, with the browser's console open. Our contact form testing checklist covers the rest, including a send that fails on the server side, which is a different story.

Then look at where people stop

With views, starts and submits counted, you can see which step loses people, and each step has different fixes.

Few starts per view means people see the form and decide not to begin. The form looks long, asks for something they won't give, or doesn't say what happens next. Put a one-line promise above it (“We reply within one working day”), cut fields you can live without, and make the first field an easy one, like a name. And if the form is three screens down the page, move it up, or put a link at the top that jumps to it.

Few submits per start means people begin and give up. That's usually a specific field: a required phone number, a captcha puzzle, an error that only appears after Send, or the wrong keyboard on a phone. Our article on why people abandon contact forms goes through those one by one, with the studies behind them. Formgong's analytics shows the last field people touched before leaving, which points straight at the field to fix.

Changes that usually help

Ask only for what you need to reply. Name, email and the message cover most contact forms; a phone field can be optional. Use the right input types, so phones open the email or number keyboard, and add autocomplete so the browser fills in what it knows. Write the button as what happens, “Send message” or “Get my quote”, rather than “Submit”. Keep your own error messages next to the field they belong to.

Put each label above its field and keep it visible; a placeholder that vanishes when typing starts is not a label. Mark the optional fields as optional, so nobody has to guess which ones they can skip. Where a format matters, say so in a short hint under the field, “e.g. +1 555 010 0123”, rather than in an error afterwards. Group related fields, contact details together and the details of the job together, so a long form reads as two short ones.

When something is wrong, keep everything the visitor typed and take them to the problem. The browser's own checks already do that: when we pressed Send with two empty required fields, Chromium put the cursor in the first one and showed its message there. If you add your own checks, run them when a field loses focus rather than on every keystroke, so nobody is told their email is invalid while they're still typing it. Lock the button while the form sends, so a second click doesn't send it twice.

Be careful with prefilled answers. Pre-selecting the most likely option saves a click, but our date test above shows a prefilled value in the wrong format simply vanished, and a wrong default that nobody changes is worse than an empty field.

Drop the captcha puzzle if you can. Every one of the 28 bot submissions in our count was caught by a hidden honeypot field that visitors never see; nobody had to pick out traffic lights. Our test of spam protection without a captcha shows what that catches and what it misses.

Watch what loads with the form. A form drawn by a third-party builder arrives after its loader script; the ones we measured were 11 to 15 KB compressed before the form itself started loading, as our form backend vs form builder comparison shows. On a slow phone, a form that appears a second late is a form some people never see.

If a form needs eight or ten questions, the usual fix is to split it into short steps with a progress line, and questions that only matter for some people can appear only after the answer that makes them relevant. Formgong doesn't build multi-page forms, so that is a job for a full form builder; for a three-field contact form, one screen is best.

And answer the question that stops people. Someone who isn't sure whether you deliver to their area or take their insurance often leaves rather than write. Formgong's AI help chat can answer from your own pages on the same site, and a few recent reviews near the form, from the same account, give a hesitant visitor the reason to send. Both sit on the same dashboard as the form.

A line under the form that says what happens to the message works the same way: it answers “who reads this, and will I end up on a mailing list?” before anyone has to ask. Formgong's form code can carry that line with a link to your privacy policy.

What happens after Send also counts

The conversion doesn't end at the click. Show a clear thank-you message or a thank-you page that says what happens next, so people don't send the form again to be sure. Then reply quickly, which only works if the submission reaches you at once. By default Formgong emails every submission as it arrives, on the free plan too, to up to three people, and to Telegram if you prefer your phone.

And judge the form by what it brings, not only by its rate. Adding a qualifying question, such as the budget or the date of the job, can lower the number of submissions and still be the better form if more of them turn into work. Count replies, bookings or sales next to the submissions.

How to know a change worked

Change one thing, wait until each version has been seen by enough people, and compare the rates with their counts, not just percentages. “31 submits from 900 views, then 44 from 920” tells you far more than “up 40%”. With fewer than a few hundred views per version, a difference of a couple of submissions is noise.

Formgong counts views, starts, submits and spam per form on every plan, once the page has the small analytics script; the hosted form page has it already. Our analytics docs explain each number.

What moved your rate?

If one change raised or lowered your form's conversion rate, tell us at support@formgong.com, with the counts before and after. Real results with their denominators help other readers more than any rule of thumb, and we will add them here.