Creating an online form takes about ten minutes. Creating one that people actually finish takes a bit longer, and almost all of the difference is in decisions made before anything is built.
This is the order that works, with the specific mistakes that cost the most at each stage.
1. Write down what you will do with each answer
Before you add a single question, write the list of things you want to know and, next to each one, what you will do differently depending on the answer. Anything with a blank next to it comes off the form.
This sounds like a productivity exercise and it is actually the highest-leverage thing in the whole process. Most forms carry two or three questions nobody has ever looked at, inherited from whatever they were copied from. Each one costs you completions and returns nothing.
The usual casualties are job title on a newsletter signup, phone number on a form nobody rings, and "how did you hear about us" on a form whose answers are never aggregated.
2. Decide the order by what it costs to answer
Cheap questions first, expensive ones last. Cheap means fast and unthreatening: a name, a choice from four options. Expensive means effort or exposure: a long explanation, a salary expectation, a phone number.
This works because of commitment. Somebody who has answered three questions is substantially more likely to answer a fourth than somebody looking at a blank form, and that effect is strong enough to carry people through questions they would have refused at the start.
The corollary is that your most important question should never be first. If you need one thing and one thing only, ask two easy ones before it.
3. Choose one page or one question at a time
A single page suits forms people are motivated to finish and want to review before submitting: applications, orders, anything with a total at the bottom. It also suits anyone who needs to skip around.
One question at a time suits forms where the length is the problem. Fifteen questions presented one at a time feel shorter than eight on a page, because nobody can see the scrollbar. It also suits phones, which is where most forms are filled in now.
The thing people get wrong is switching for aesthetic reasons. Pick based on whether seeing the whole form would help or frighten the person filling it in.
4. Use logic instead of asking everyone everything
The single biggest improvement available to most forms is conditional logic: showing a question only to the people it applies to. A form that asks fifteen questions but shows any given person six feels like a six-question form, because it is.
The pattern is a routing question near the top, then branches. "What is this about" with four options, and four different sets of follow-ups. Everybody answers one question more than a specialised form would need, and nobody answers nine they do not.
Use it for the awkward cases too. "Do you need an invoice" followed by billing details only for the people who said yes is better than six accounting fields nobody else has to scroll past.
5. Write the questions as questions
"Company" is a label on a database column. "Which company do you work for?" is something a person can answer. The second gets better data, consistently, because it removes the moment where somebody has to work out what you meant.
Say what optional means. A field marked optional with no explanation reads as a trap. "Optional, only so we can send you a copy" gets filled in far more often than the same field unlabelled.
And give an example where the format is ambiguous. Date fields, reference numbers and anything with a country-specific shape are all cheaper to get right with a placeholder than to clean up afterwards.
6. Test it the way it will actually be used
Fill it in on your phone, outdoors, in a hurry. That is the real condition for most forms and it exposes problems that a desktop preview never will: tap targets too close together, a date picker that fights you, a text box that pushes the submit button off the screen.
Then fill it in wrongly on purpose. Leave required fields blank, type letters into the number field, submit twice. What you are looking for is whether the error message tells you what to fix and whether it appears next to the thing that is wrong.
Finally, have somebody who did not build it try. Say nothing while they do. Every place they hesitate is a question that needs rewording, and you will want to explain rather than write it down. Write it down.
Fill it in on your phone, outdoors, in a hurry. That is the real condition, and it finds problems a desktop preview never will.
7. Decide what happens after submit
The page after submit is the most wasted space in forms. "Thank you, your response has been recorded" tells somebody nothing they wanted to know. What they want is when they will hear back, what happens next, and whether they need to do anything.
Confirm the detail back to them. For an order, the order. For a booking, the time. For an application, the deadline you are working to. This single change removes a large share of "did my form go through" emails.
And give them something to do. A link to the thing they were reading, a calendar invitation, a document they wanted. The person who has just filled in your form is the most engaged visitor you will have all day.
The mistakes that cost the most
In rough order of damage done:
- Asking for a phone number you will never ring. One of the most expensive fields in conversion terms.
- A required field that should be optional. Each one turns a partial response into no response.
- Validation that rejects legitimate input. Phone numbers, names with apostrophes and non-UK postcodes are the classic offenders.
- No indication of length. People abandon unknown forms far more readily than long ones they were warned about.
- Error messages at the top of the page rather than next to the field.
- Asking the hardest question first.
- A generic thank-you page that says nothing about what happens next.
Starting from something
Starting from a blank form means making every one of these decisions yourself, which is why most people start from a template and cut. Pick the closest one, remove what you cannot act on, and reorder so the cheap questions come first.
That takes about ten minutes and produces a better form than an hour from scratch, because the structural decisions are already made and the only work left is the part that is specific to you.
What to measure once it is live
A form you never look at is a form that quietly stops working. Three numbers tell you almost everything.
Completion rate is how many people who started also finished. Under about sixty percent on a short form means something is wrong rather than that your audience is fickle.
Drop-off by question is the one that tells you what. Somewhere in most forms there is a single question where a disproportionate share of people leave, and it is almost never the one the team guessed. It is usually a phone number, a required field that should not be, or a question people do not understand.
Time to complete against what you told them. If you said two minutes and the median is six, the problem is the promise rather than the form.
Accessibility, briefly and properly
Most of this is free if you do it while building and expensive to retrofit.
- Every field needs a visible label, not just a placeholder. Placeholders disappear the moment somebody types, which leaves people re-reading a form they have already half filled in.
- Errors must be announced, not only coloured. Somebody using a screen reader cannot see that a border turned red.
- The form has to work with a keyboard alone, in a sensible order. Tab through it once and see where focus goes.
- Do not rely on colour alone to mean anything. Required, error and success all need a word or an icon as well.
- Give people time. Anything that times out is a barrier to somebody who reads slowly or is interrupted.