Every team has the conversation about shortening the form, and it usually arrives with a quotation: remove one field and conversion rises by some percentage. The direction is broadly right — fewer fields converting better is a widely observed tendency. But that sentence cannot tell you which field to remove, and it certainly cannot tell you what would happen on your site.
A more useful question: what does this field do to earn its place?
Four questions per field
The third question is the strongest one. Anything you do not need yet can be collected later. Our own application form is built on exactly that judgement: server access details are not requested at application. They are collected in the customer portal once payment is confirmed, because asking for server credentials up front creates resistance and the information is not needed at that moment anyway.
Friction that is not about field count
Plenty of friction can be removed without deleting a single field, and most of these changes cost almost nothing.
In particular, do not use placeholders as labels. The guidance disappears the moment typing starts, leaving the customer no way to confirm what they were answering. Labels belong somewhere permanently visible.
What happens after submission is part of the form too. If the confirmation screen and the auto-reply do not arrive, the customer cannot tell whether it worked, so they submit again or leave. Before touching the fields, send yourself a test submission and check both the inbox and the spam folder.
Confirming a change
Before changing anything, find out where people stop. Form analytics will tell you the last field someone was in; failing that, at minimum record form views against submissions per page. Without a baseline there is no way to say whether a change was an improvement.
Then change one thing at a time. Remove a field while also rewriting the copy and recolouring the button and you will never learn which one did the work. And decide your criteria — sample and duration — before you start, which is exactly the next part. Related writing continues in the SEO archive.
Next part
An A/B test does not begin with a start button; it begins with two decisions. The next part covers both, and what happens when you skip them.