Most businesses measure their contact or purchase form with a single figure: how many submissions it received over a given period. That is a useful but incomplete metric, because it says nothing about everyone who started filling it in and abandoned it halfway through, or at what exact point they left. Without that detail, it is impossible to know whether the problem is the first field, a phone number requested with no explanation, or a submit button that does not display well on mobile.
The difference between form conversion rate and field-by-field abandonment rate
The overall conversion rate (submissions divided by page visits) is the starting point, but the truly actionable data is the field-by-field abandonment rate: how many people start interacting with the form, how many reach the second field, the third, and so on up to submission. This field-by-field view is what reveals exactly where the biggest leak happens, instead of leaving it to blind guesswork.
How this is actually measured
Specialised form analytics tools (some built into heatmap platforms, others standalone) record when a user clicks into or enters each field, how much time they spend there, and whether they complete it or abandon it unfilled. With Google Tag Manager you can also set up more hands-on tracking, firing an event every time the user leaves a field (a "blur" event, technically) to later reconstruct which field caused the most abandonment.
The fields that cause the most leaks, by a wide margin
Some patterns repeat across the vast majority of forms analysed: the phone number raises more doubts than email, especially if it is not explained why it is being asked for or if it looks like it will be used for unwanted sales calls; full address fields (street, number, floor, postcode in separate boxes) generate far more abandonment than a single address field with autocomplete; and any field asking for information the person does not understand why you need (a company name on a private individual's contact form, for example) slows down conversion without adding real business value in most cases.
The mistake of asking for too much, "just in case"
It is common for a form to include fields that nobody actually uses afterwards to make decisions (how did you hear about us?, job title, number of employees) simply because "it would be nice to have that data". Every additional field, however small it looks, measurably reduces conversion rate. The question to ask before adding any field is not "would it be useful to know this?" but "are we actually going to use this information for something specific, and does it outweigh the conversion cost of asking for it?"
Long forms split into steps: do they help or hurt?
Splitting a long form into several steps (with a visible progress bar) usually reduces perceived effort and improves completion rate compared to the same form shown all at once on one long screen, as long as each individual step is short and the progress bar makes clear how much is left. The risk lies in the first step: if it creates friction or doubt before the person has already invested some effort, abandonment at that first step can even be higher than with a single-screen form.
Error messages: the most underrated source of abandonment
A field showing an unclear error message ("invalid format", without saying what format is expected) or one that validates at the wrong moment (flagging an error while the person is still typing, before they finish) generates enough frustration that many people simply abandon instead of trying to fix it. Validating only once the field is left (not while typing) and with specific messages ("phone number must have 9 digits") notably reduces this kind of abandonment.
Forms on mobile: their own set of rules
On mobile, field size, the type of keyboard that pops up (numeric for phone, with a visible @ key for email) and the position of the submit button relative to the on-screen keyboard (which can hide it if not accounted for) create abandonment patterns that do not exist on desktop. It is worth analysing form performance on mobile separately from desktop, because mixing both almost always hides problems specific to one of the two.
A concrete case: the form that lost half its people on a single field
A real estate agency had a rental information request form with seven fields, and a mediocre completion rate nobody could explain. On implementing field-by-field analytics, it became clear almost half of the people who started filling in the form abandoned right at the "monthly income" field, a data point the agency asked for out of habit but in practice almost never used for anything at that first-contact stage. Removing that field from the initial form, and moving it to a later stage of the conversation (once direct contact and trust had already been established), recovered a very significant share of the requests that had previously been lost with nobody knowing why.
The submit button is also part of the analysis
A detail often overlooked: the submit button's own text influences the form's final conversion. A generic button that says "Submit" communicates nothing about what happens next, while more specific text ("Request information, no obligation" or "Book my free appointment") reduces last-minute uncertainty right at the decisive moment, and usually measurably improves the final conversion rate, a change that takes minutes to implement and is rarely tested deliberately.
What to do with submissions that fail from a technical error, not abandonment
Not every conversion lost on a form is down to someone deciding not to fill it in, some can be down to silent technical errors: a submission that fails from a connection issue, a field with badly coded validation rejecting correct data, or a character limit too short for a long address. Reviewing the form's own error logs (if the platform offers them) and not just behavioural analytics helps tell apart "didn't want to" from "couldn't", two problems with completely different solutions.
Step by step: setting up field-by-field tracking on a form
To implement this kind of analytics without needing an expensive tool, the minimum viable process is this: first, decide which form to audit (always start with the highest business-value one, not the easiest to set up). Second, in Google Tag Manager, create a "click" or "form" trigger for each field on the form, configured to fire when the user leaves the field (a "blur" event). Third, create a Google Analytics event tag tied to each trigger, naming the event clearly ("form_phone_field_completed", for example). Fourth, publish the container and wait for at least two weeks of real traffic. Fifth, in Analytics, build a funnel or exploration showing how many people complete each field in order, through to final submission. Sixth, identify the field with the biggest percentage drop compared to the previous one: that is the first candidate to review, whether by removing it, repositioning it, or explaining more clearly why it is being asked for. This process, while it needs some initial technical setup, only needs to be built once and keeps generating useful data continuously with no further maintenance.
Frequently asked questions
How many fields should a contact form have at most?
There is no fixed number valid for every case, but as a general rule, fewer fields means higher conversion; the key question is how many of those fields are truly essential to take the first step in the relationship with that person, leaving the rest of the information to be gathered later in the conversation.
Is it better to make the phone number field required or optional?
It depends on the business: if the first real contact happens by phone, it can be justified as required, briefly explaining why it is asked for. If the business can work perfectly well with the first contact by email, making it optional usually improves conversion rate without losing too many quality leads.
Is it worth using address autocomplete with an API instead of manual fields?
Yes, in most cases it reduces fill-in time and typos, and notably improves the mobile experience, where typing a full address by hand is especially uncomfortable.
How do I know if my form's problem is the design or the information it asks for?
Field-by-field analytics tells them apart fairly well: if abandonment concentrates on one specific field regardless of its position, it is usually a problem of what is asked or how it is explained; if abandonment grows steadily as the form progresses with no clear spike at any field, it usually points more to general fatigue from the form's length.
Does it make sense to run A/B tests on forms with low traffic?
With low traffic, it is better to directly apply improvements based on already widely validated best practices (reducing fields, validating errors properly, splitting into steps if it is long) rather than waiting to accumulate enough traffic for a formal, reliable A/B test.
Should I analyse forms coming from paid advertising separately from those coming from organic traffic?
Yes, because intent and prior familiarity with the business usually differ between the two groups, which can create different abandonment patterns that a combined view might hide.