Fix the quote form before you redesign the site.

On 10 November 2026 we opened 50 quote and contact forms on the websites of B2B companies in the UK, the US, the EU, the Gulf and Australia. We counted the fields and read the labels. On 49 of them we typed one deliberate mistake and pressed Send, with the browser set to block every outgoing request after the click, so no enquiry was submitted. The median form had 7 fields. Of the 50, 42 asked for a phone number and 26 required one. None of the 42 gave a reason beside the field.

If you are about to pay for a redesign to win more enquiries, start with the form. It is the step where a visitor becomes a lead, and you can change it on its own and compare results before and after.

A redesign changes too much to measure

A redesign changes layout, copy, navigation, speed and usually the form, all at once. If enquiries rise afterwards, nobody can say which change did it. Redesign uplift figures are easy to find. The one we traced in detail was run by a company that sells redesigns. It changed forms, copy, navigation and speed together, and it states that “Revamp conducted and funded this internal study”.^[1]

Nielsen Norman Group’s advice is to improve a site in steps and “only resort to a complete redesign if the data tells you to do so.”^[2] Start collecting that data at the form.

What we found in 50 quote forms

We took two companies from each of five sectors in each region: IT and software services, freight and logistics, industrial suppliers, engineering and construction consultancies, and B2B marketing and design agencies. We left out directories, marketplaces and companies that looked larger than about 1,000 staff. The method is at the end of this post.

Submit buttons were rarely the problem. All but one of the 50 submit buttons measured “at least 24 by 24 CSS pixels” at desktop width, the minimum target size in WCAG 2.2 success criterion 2.5.8 (Level AA).^[3] On 19 forms at least one field had no label linked in code, and 39 gave no reply time near the form.

Six fixes, graded by their evidence

Each fix below carries a grade for the evidence behind it. A standard is a published accessibility specification, such as WCAG, that a site can be built to; which standard a law points to varies by country. Research means usability research or a peer-reviewed study. Guidance is advice from a design system or a standards working group. A vendor case is a result reported by the company that ran it, without independent review. Where good sources disagree, we say so.

1. Cut to the fields your first reply needs

Baymard Institute’s checkout research found that, in general, “the number of form fields in a checkout impacts overall usability far more than the number of steps.”^[4] Nielsen Norman Group puts the risk plainly: “Every extra field increases the chance they’ll give up.”^[5]

The closest B2B number we found is one company’s own A/B test, reported by MarketingSherpa. Removing two qualification fields from a trial sign-up form moved it from “120 submissions out of 5,000 visits (2.4% conversion rate)” to “172 submissions out of 5,000 visits (3.44% conversion rate)”.^[6] That is one form at one company, and the figures are the company’s own. We found no independent study that sets a right number of fields for a B2B quote form.

Ask for what your first reply needs. Budget, company size and how they heard of you can wait for the call. In our audit, 11 of 50 forms had 10 or more fields, and 11 required every field they showed.

2. Say why you want the phone number

Baymard’s checkout research advises that “sites should clearly explain why a phone number is required, make the field optional, or remove it entirely.”^[7] Of the 42 forms in our audit that asked for a phone number, 26 required it. None of the 42 gave a reason in the field’s label, placeholder or hint.

If you will call, say so and say when. If you reply by email first, make the phone field optional.

3. Put a label on every field, and mark what is required

A placeholder disappears when the visitor starts typing. Nielsen Norman Group warns that “Without labels, users cannot check their work before submitting a form.”^[8] WCAG 2.2 success criterion 3.3.2 (Level A) states: “Labels or instructions are provided when content requires user input.”^[9] Our audit counted something narrower: whether each field had a label linked in code, and whether label text sat beside it. 31 of 50 forms linked a label to every field in code, and 14 had at least one field labelled only by placeholder text. Across all 401 fields, 75% had a linked label and 16% had only a placeholder. Of the other 35 fields, 23 had label text beside the field, which leaves 12 with no visible label at all.

Required marks are where good sources disagree. Nielsen Norman Group recommends marking required fields and notes that “Only marking optional fields makes it difficult for people to fill out the form.”^[10] The GOV.UK Design System marks optional fields instead: “Never mark mandatory fields with asterisks.”^[11] Pick one rule, say which at the top of the form and apply it to every field. In our audit, 15 of the 50 forms had no required marking at all.

4. Describe the error next to the field

WCAG 2.2 success criterion 3.3.1 (Level A) covers errors a form detects itself: “the item that is in error is identified and the error is described to the user in text.”^[12] The criterion does not say where the message goes, so placing it next to the field is our recommendation. Our test typed “not-an-email” into the email field, left the rest empty and pressed Send. One form stopped responding before the test, so the chart counts 49.

Within two seconds of the click, 19 of the 49 stopped at the browser’s own warning bubble. 15 showed an error message in the page, and 11 of those placed every message at its field. Another 9 tried to send and showed nothing before the request was blocked, so any reply from their server would show only if the request got through, which we never allowed. On one, the click did not register. The other 5 showed a red border alone, disabled the button, raised a pop-up or did nothing visible. In all, 16 of the 49 made a request that could have been a submission, and the browser blocked each one.

Sources also split on when to validate. Baymard’s e-commerce testing found that “the validity of each field input should be checked when the user leaves the field”.^[13] GOV.UK advises the opposite: “Do not validate when the user moves away from a field.”^[14] A 2009 A List Apart article reported a usability test of “22 average users on six variations of a typical web registration form.”^[15] On a 5-field quote form, checking on Send with a clear message at each field is the simpler choice to build and test.

5. Say when you will reply

The GOV.UK Design System says confirmation pages reassure users and help them “understand what to expect next.”^[16] A quote form can make that promise before the visitor presses Send. In our audit, 15 of the 50 forms said anything near the form about a reply, and 11 gave a time, such as within one business day. Promise a time your team keeps, and name the button for what it does: Send enquiry, not SUBMIT.

6. Try a honeypot before a CAPTCHA

CAPTCHAs cost people time. A 2023 study in which 1,400 participants solved them reported that “the average solving time in our study ranged from 3.6 to 42.7 seconds per captcha”, depending on the type.^[17] A W3C draft note on CAPTCHA accessibility advises that users “should not be forced beyond what is strictly necessary to keep a site secure” and that “if a honeypot suffices, use a honeypot until evidence of robotic attacks dictates something else.”^[18]

A CAPTCHA loaded on 22 of the 50 pages, and 11 showed the box as soon as the page opened. On 13, a likely honeypot was the only protection we could see. Start with a honeypot and a server-side spam filter, hide the honeypot from screen readers, check that browser autofill does not fill it, and add a challenge if spam still gets through. The evidence here covers what CAPTCHAs cost visitors. It does not rank spam filters, so watch your spam rate after the change.

A form statistic, traced

Articles about forms often say that cutting fields lifts conversions by 120%. The earliest write-up we found is below.

Ask the same three questions of any redesign uplift figure: who ran the study, who paid for it, and what else changed at the same time.

Redesign when the data says so

Fix the form, then measure for four weeks. Track qualified leads per 100 visits to the form page, and compare with the four weeks before, keeping both periods free of campaigns and holidays. If the form gets few visits, extend the window until there are enough enquiries to compare. Then look at how many visitors reach the form at all. If the form now converts well and too few people reach it, the pages before it are the case for a redesign, and you have numbers to brief it with.

How we checked

We split the sample into 25 groups, five regions by five sectors, and took the first two qualifying companies from the results for each group: 25 searches, plus 3 follow-ups where a group needed more, run on 10 November 2026. EU searches covered the Netherlands, Germany, Poland, Ireland and Spain. Gulf searches covered the UAE, Saudi Arabia and Qatar. We skipped directories, marketplaces, ranking articles, large enterprises and sites that blocked automated browsers. A page qualified if its form had at least three visible fields.

Each form was read once, on the search result page or one link from it, in headless Chrome, a browser driven by code that identified itself as such, at desktop width and on the first step of multi-step forms. For the error test, every network request after the click was blocked inside the browser, and a check before each click confirmed the block, so no enquiry was sent. We watched each form for two seconds after the click. We read the phone field’s label, placeholder and hint for a reason, and the text around each form for a reply promise. Companies are not named.

Verne’s designers and growth engineers work on your existing site, apply the fixes that fit your form and agree how each change will be measured before it goes live. The code is handed over at full delivery. To talk it through first, book a diagnostic.