Form fields are a conversion tax - and most sites are overpaying.

Every field is a tax - and most sites overpay.

The median form converts at roughly 17.3%. That means for every 100 people who actually start typing, 83 leave without hitting submit. Not because they weren't interested - they were interested enough to start - but because the form made finishing harder than leaving.

Across the 1,284 pages Levri has analysed, the pattern is almost boringly consistent: sites ask for too much, validate too late, and punish people for trying. And the damage is invisible in most analytics - you see the page-level drop-off and shrug, because the dashboard never shows you that a single required phone field is quietly responsible for 37% of all form abandonment (Klaviyo, 2024).

The good news: form leaks are the most fixable thing on the page. No redesign, no new traffic. Here are the six patterns we see most, and what to ship for each.

The six patterns that move completion.

1. The field-count cliff - conversion collapses between 5 and 7 fields.
2. The required phone trap - one field drives a third of all abandonment.
3. Single-page form walls - long forms with no steps feel like work.
4. Missing inline validation - the submit-scroll-fix-scroll death spiral.
5. Broken autofill - markup that actively blocks the browser from helping.
6. Error messages that punish - "Invalid input" instead of a way forward.

Most forms we scan are running three or four of these at once. Fix the top two and you're usually looking at a double-digit completion lift before touching anything else.

#1 - The field-count cliff.

Every field you add costs conversion - but the cost isn't linear. There's a cliff between five and seven fields where each new input costs roughly twice what the last one did. A three-field form converts around 23%; a five-field form drops to ~17%; add two more and you're at 11.4% - a third of your completions gone to two extra boxes (Digital Applied, 2026).

The deeper version of this - what the minimum viable signup actually looks like - is in reduce signup friction.

What we see on scans

• Lead-gen forms with 8-12 fields - company name, job title, phone - all marked required.
• Checkout forms averaging ~15 fields where 6-8 is plenty for guest checkout.
• A "company size" dropdown driving 22% single-field abandonment on its own.
• Marketing-preference checkboxes crammed into the conversion flow.

What to ship

• Audit every field against what you need at this stage - not what the sales team wishes they had.
• Cut company name, fax, address line 2, and password confirmation today.
• Move nice-to-haves to a post-conversion progressive profile.
• Target 5 fields or fewer for lead-gen, 6-8 for guest checkout.

Typical lift: +10% to +20% on mobile, +7% to +14% on desktop. Impact: high - the single highest-leverage change in form optimisation.

#2 - The required phone trap.

Phone is the most abandoned required field there is - Klaviyo's 2024 audit pins it at 37% of all form abandonment, more than any other single field. The objection isn't typing digits; it's what the digits imply: spam calls, follow-ups, a relationship they didn't agree to. Marking phone required tells people you value your outreach over their comfort.

What we see on scans

• Phone required on standard ecommerce checkout with no delivery justification.
• SaaS demo forms with required phone, triggering sales calls people didn't expect.
• No explanation anywhere for why the number is needed.
• Digital-product checkouts asking for a phone number when nothing ships.

What to ship

• Make phone optional - this one change lifts checkout completion 5-8% (Baymard Institute).
• If you genuinely need it, add one line: "We'll only text you delivery updates."
• For digital products, remove it entirely - that's another 2-4%.
• For returning customers, auto-populate it and skip the ask.

Typical lift: +5% to +8% switching from required to optional. Impact: high - wildly disproportionate to the effort (one attribute).

#3 - Single-page form walls.

A nine-field form on one page looks like work. The same nine fields across three short steps looks manageable. The content is identical; the perception isn't. Multi-step forms with a progress indicator convert ~14% higher on average, and 21% higher on B2B lead-gen forms with six-plus fields (Digital Applied, 2026). The mechanism is foot-in-the-door: once someone finishes step one, sunk cost carries them forward.

What we see on scans

• 8-12 field forms rendered as a single scrollable block.
• Multi-step forms with no progress indicator.
• Step one asking for the hardest information (budget, timeline) before the easiest (name, email).
• Mobile forms where submit sits three scroll-lengths below the first field.

What to ship

• Split any form with five-plus fields into 2-3 steps.
• Add a visible "Step 2 of 3" - that alone adds another 9-15% on top of the multi-step gain.
• Put the lowest-friction fields first; save sensitive or complex ones for later.
• Keep each step to 2-4 fields.

Typical lift: +14% to +21% for six-plus-field forms; roughly neutral under four fields. Impact: high - especially B2B lead-gen and demo flows.

#4 - Missing inline validation.

The form experience on most sites: fill in eight fields, hit submit, watch the page reload with three red errors at the top, scroll up, fix, scroll back, submit again. That loop drives 28% of multi-error abandonment. Inline validation - pass/fail feedback as you leave each field - lifts completion 5-13%, and 11-13% on forms with six-plus fields (Baymard, 2024). Small errors caught immediately feel like help; the same errors caught after submit feel like punishment.

What we see on scans

• All errors surfaced on submit only, never field-level.
• Generic banners ("Please fix the errors below") with no field named.
• Email fields that accept input silently, then reject it after submit.
• Card fields rejecting spaces and hyphens - a top hidden conversion killer.

What to ship

• Add on-blur validation to every required field - check format the moment focus leaves.
• Show green ticks on valid fields, not just red on invalid - positive reinforcement matters.
• Accept format variations: spaces in card numbers, hyphens in phone, dots in postcodes.
• Put error messages inline, directly under the field - not in a banner.

Typical lift: +5% to +13%, high end on longer forms. Impact: medium-high - front-end work, but it pays on every submission.

#5 - Broken autofill.

Browser autofill cuts completion time 30-40% and lifts mobile conversion 12-18% - yet 39% of ecommerce sites break it with the wrong field attributes (BuildGrowScale, 2026). The fix is almost embarrassing: use correct autocomplete attributes. The browser already knows the user's name, email, address, and card. When the form cooperates, 90 seconds of typing collapses into 10 seconds of tapping. The mobile stakes here are the same ones in mobile checkout optimization.

What we see on scans

• Inputs with autocomplete="off" - actively blocking the browser from helping.
• Non-standard names (field_7, custom_email_input) that autofill can't parse.
• Address captured as one freetext box autofill can't populate.
• Checkout where autofill works on 2 of 8 fields - a frustrating partial fill.

What to ship

• Add correct attributes: autocomplete="given-name", ="email", ="street-address", and so on.
• Remove autocomplete="off" unless there's a real security reason (passwords aside).
• Use structured address fields, not a single textarea - it enables autofill and cuts errors.
• Default country from IP, and pre-select the most popular shipping method.

Typical lift: +12% to +18% on mobile, +6% to +10% on desktop. Impact: medium - mostly markup, but the mobile win is real.

#6 - Error messages that punish.

"Invalid input." Two words that cost 14% of completions versus specific, helpful messages (Baymard, 2024). When a form rejects input without explaining why, people don't troubleshoot - they leave. An error message isn't a validation detail; it's a micro-conversation where you either help the user forward or slam the door.

What we see on scans

• "Please enter a valid email" with no hint what's wrong with john@company.
• "Error in field" with nothing highlighted.
• Date fields rejecting DD/MM/YYYY when that's the user's locale.
• Submissions that clear every field on error, forcing a restart.

What to ship

• Name the problem and show the fix: "Email needs an ending like @gmail.com" beats "Invalid email."
• Preserve all entered data on failure - never clear the form.
• Accept multiple date and phone formats, or use input masks that enforce them visually.
• Auto-format as people type: card numbers in groups of four, masked dates, formatted phone.

Typical lift: +8% to +14% replacing vague errors with specific ones. Impact: medium - low effort, meaningful gain.

How Levri spots all six in 60 seconds.

Paste a URL. Levri finds every form on the page and flags field-level friction - missing autofill attributes, excess required fields, absent inline validation - then ranks each by revenue impact so you know what to fix first.

You paste the URL, and you get your fixes - a ranked list, each issue priced in $/mo, with a written hypothesis, a variant-B suggestion, and an expected lift range. No install, no tracking script, no "connect your analytics."

Fix these first.

In the order we'd ship - each is hours, not a sprint:

1. Cut to five fields or fewer (highest leverage, lowest effort).
2. Make phone optional (one attribute, 5-8%).
3. Add inline validation (front-end work, 5-13% back).
4. Split long forms into steps (14-21% on six-plus-field forms).
5. Fix autofill attributes (markup-only, big mobile win).
6. Rewrite error messages (small effort, 8-14%).

The same field discipline shows up at the payment step in the checkout leak taxonomy. Ship three of these and measure by Friday - if nothing moves in two weeks, you're fixing the wrong leaks, which is exactly what Levri is built to diagnose.