What Happens After Someone Clicks Your Submit Button
A roofing client called us because his contact form was broken. It was not broken. Over six weeks it had received 41 submissions and 11 of those were duplicates, the same name and phone number sent two or three times within a minute.

A roofing client called us because his contact form was broken. It was not broken. Over six weeks it had received 41 submissions and 11 of those were duplicates, the same name and phone number sent two or three times within a minute. People were clicking submit, seeing nothing change, assuming it had failed, and clicking again. Two of them had typed "did this send?" into the message field on the second attempt.
That form validated correctly, sent the email reliably, and still lost the trust of a quarter of the people who used it. The failure was entirely in the second and a half after the click, which is the part of a form nobody designs because it does not show up in a mockup.
Games Solved This Thirty Years Ago
I drew HUD elements for two indie studios before this job, and the single most argued-about thing in any combat system is hit feedback. When your attack lands you get a sound, a flash, a number, a tiny freeze of both characters, and a controller rumble. Five channels of feedback for one event. Nobody considers this excessive, because an attack with no feedback feels like the game ignored you.
Fighting games take it furthest with hit-stop, a deliberate freeze of two to five frames on contact. It makes the game less responsive, on purpose, because the pause is what makes the hit feel like it connected. A designer chose to add a 50 millisecond delay to improve how an action felt.
A submit button is an attack that lands in a database. It gets no sound, no flash, no number, and usually no acknowledgement at all until a page reloads. The gap between how carefully a studio treats a sword swing and how casually a web team treats a lead submission is, honestly, difficult to defend.
The Four Time Windows and What Each One Needs
Response times group into bands that have been stable in the research since the 1960s and match what I saw in playtests. The table below is what we build to now.
| Delay after the click | How it feels | What to show |
|---|---|---|
| Under 100 ms | Instant, caused by you | Button press state is enough |
| 100 ms to 1 second | Noticeable but connected | Button label changes, button locks |
| 1 to 5 seconds | Waiting, attention drifting | Spinner inside the button plus changed label |
| Over 5 seconds | Suspicion that it failed | Progress text that names the step, and a timeout path |
The roofing form sat in that third band at around 2.4 seconds, measured on a mid-range Android on a rural connection, and showed nothing at all. On the designer's laptop it returned in 300 milliseconds, which is why nobody caught it. Test submit timing on a throttled connection, not on your desk.
The label change matters more than the spinner. "Sending your request" tells the visitor what is happening. A bare spinner tells them something is happening, which they already assumed. And lock the button on the first click, because that alone would have removed 11 duplicate leads from my client's inbox.
Why the Thank You Page Is Usually the Wrong Choice
Standard advice says redirect to a dedicated thank you page, because that is where you fire conversion tracking and where you can place the next offer. Every agency including ours did this by default for years. I now think it is wrong for most small business forms.
The redirect costs you the context. A visitor who filled a form in the middle of your pricing page is now on a page that has nothing to do with what they were reading, and the back button takes them to an empty form. On mobile, over a slow connection, a redirect also means a second full page load during the exact moment you want to look competent.
The tracking argument has not been true for a while. Event-based analytics fires on the submit response perfectly well, and a dedicated URL is a convenience for whoever set up the goal, not a benefit to the visitor.
So we replace the form in place with a success block, same width, same position, so the page does not jump. It carries a specific confirmation, a realistic response time, and a named next step. Here is the version that beat our old thank you page on a kitchen fitter's site by about 14 percent on follow-up engagement, measured over four months on roughly 200 submissions, which is a small enough sample that I would not bet the farm on it:
Thanks Dale, your quote request is in. Marcus reads these himself and usually replies within one working day, two if you have sent it over a weekend. If it is urgent, his mobile is on the contact page.
Name the person. Give the real number of hours. Offer the faster channel for people who cannot wait. That last line feels like it cannibalises your nice tidy form workflow, and it does, and the people who use it are the ones who were going to call a competitor instead.
I will grant one exception. If the form starts a genuine multi-stage process, such as a booking that needs payment or a file upload, a separate page is right because the next stage needs the room. The in-place success block is for the common case of one form, one message, one reply.
Failure States, Which Are the Ones Nobody Builds
Ask a designer for the error state of a form and you will usually get red text under an empty required field. That is validation, which happens before the click. The states that actually damage trust happen after it.
The send fails. Your mail service is down, the API times out, something returns a 500. If your form shows a success message anyway, which plenty of them do because the success handler is the only handler anyone wrote, you have just told someone you received a message you did not. We now show the typed content back, preserved exactly, with a line saying the send did not go through and here is the phone number and email address to use instead. Losing a form submission is bad. Losing it silently is much worse.
The visitor leaves mid-form and comes back. Game designers would call this a save point, and a form that has forgotten everything is a game that deleted your progress. Storing field values locally as the visitor types costs very little and is noticed immediately by anyone who has fat-fingered a back gesture on a phone.
The double submit from a slow connection, which is the one that started all this. Lock the button, and on the server side deduplicate on the same email and message within a short window. Both, not one.
And the quiet one: the reply never comes. A beautifully designed confirmation that promises one working day and delivers nothing in nine days has done more harm than a plain one. The confirmation state is a promise your client has to keep, so write the number of hours they will actually hit rather than the number they would like to hit. The rest of the form, from field order to field count, is in designing a contact form people actually finish.
A Short Audit You Can Run This Week
Throttle your connection in dev tools to a slow 3G profile and submit your own form on a phone. Watch what the button does. Count the seconds before anything changes. Then submit it twice quickly and see whether two records arrive. Then break it on purpose, by taking the form action offline, and see what the visitor is told.
Four tests, fifteen minutes, and in my experience three of the four fail on a site that was built by someone competent, because the success path is the only path anyone exercises. The words on the button are the other half of this and they deserve the same care, which is the argument in writing button copy that tells people what happens next. More of this sort of thing lives in the conversion notes.
FAQ
Should the submit button be disabled while sending?
Lock it against further clicks, but change the label at the same time so it reads as working rather than broken. A greyed-out button with the same label on it is the most common version of this and it looks like a bug.
How long should I promise to reply in?
Whatever the business actually does, plus a margin. If replies realistically take a day and a half, say two working days. Beating your own stated time is free goodwill and missing it is the fastest way to be forgotten.
Do I still need a thank you page for conversion tracking?
No. Fire an analytics event when the submit response succeeds. The one practical reason to keep a thank you URL is a third-party ad platform that only accepts a destination URL as a conversion, and even most of those now take events.
Is an animation on success worth adding?
A short one, under 300 milliseconds, that draws the eye to the confirmation text. Anything longer delays the information the visitor came for. Games earn their long celebration animations because the player has nowhere else to be, and your visitor does.
More in Calls to Action
Calls to Action
Multi-Step Form Progress Indicators That Keep People Going
A garden design studio came to us with a quote request form that had grown to five steps: project type, garden size, budget,...
Calls to Action
Exit Intent Popup Design That Does Not Annoy Visitors
Most visitors leave a small business website without doing anything, and often they were close.
Calls to Action
Split Testing a Small Business Website With Low Traffic
You changed the button colour, ran a split testing tool for two weeks, and got a result that flipped back and forth every few...