A B2B scheduling software company came to us with a demo problem: 1,480 monthly visitors to the demo page and 31 bookings. A 2.1% conversion rate, in a category where 4% is normal.
Thirty days later the rate was 4.4% and bookings had doubled to 64 a month. The core change was not a redesign, new copy, or more traffic. We deleted one form field — then let an agent replace the data the field used to collect.
Here is the whole case, numbers included, client anonymized as always.
The setup: seven fields and a stalled funnel
The demo form asked for seven things: first name, last name, work email, company, company size, phone number, and a free-text box titled 'what are you hoping to solve.' Classic lead-gen thinking, where every field exists because sales wanted the context. Traffic was healthy and stable at around 1,480 visits a month, the page loaded fast, and the copy tested fine in every review. The funnel died at the form itself: 68% of the people who started it never finished.
Sales loved the form. Buyers, evidently, did not.
The client's instinct was a redesign. The data argued for something much smaller.
Context for scale: the average contract here runs about $14,000 a year, so each demo is worth real money in expectation. The form was quietly discarding a third of them every month, with zero visibility to anyone.
What the field-level data showed
We instrumented the form per field before touching anything: focus events, time in field, and the last field touched before abandonment. Two weeks and roughly 640 form starts told a clean story. The phone field was the last field touched in 44% of abandonments, and time-in-field tripled when people reached it — the hesitation was visible in the timestamps. The free-text box accounted for another 19%. Nobody abandoned on name or email. The form did not have a friction problem. Two fields did.
This is the step most teams skip. Page analytics said the form underperformed. Field analytics said which field, specifically.
Two weeks of instrumentation cost nothing and pointed the entire engagement.
The tooling was deliberately ordinary: a 40-line event script and a shared spreadsheet. Nothing about the method needs an enterprise analytics contract. It needs someone willing to look one level deeper than the page view.
The one deletion that mattered
We deleted the phone field first and changed nothing else, to keep the test readable. Completion rate on form starts went from 32% to 51% inside two weeks, and bookings jumped from 31 to 47 a month on flat traffic. Then we removed the free-text box and made company size optional, and completion settled at 61%. The phone field alone had been suppressing roughly a third of all demo bookings, month after month, for as long as the form had existed.
People will give a stranger their work email. A phone number reads as an invitation to be chased.
The free-text box was a different failure: it asked buyers to write sales copy for the vendor. At 11pm, nobody wants homework.
Order mattered as much as the deletions. One change per two-week window kept every effect attributable, which is what made the client trust the next cut.
Where the missing data came from instead
Sales still needed the context those fields carried, so an enrichment agent now fills the gap. On every booking it takes the work email domain, pulls company size, industry, and tech stack from public sources, drafts a one-paragraph brief, and posts it to the sales channel within four minutes of the booking. Accuracy against what buyers used to type ran 91% on company size. That is not a typo — the agent beat the humans, because buyers guessed and the agent looks it up.
The phone number sales wanted for confirmations comes later. The confirmation email asks for it once the meeting exists, and 72% of bookers provide it.
Sequencing is the entire trick: ask after commitment, not before it.
Cost of the agent in production: about $9 a month in model and data calls. Four minutes also beat the old workflow, where a rep skimmed the free-text answer moments before the call started.
The numbers after 30 days
Before: 1,480 monthly demo-page visits, 31 bookings, 2.1% conversion, 61% show rate. After 30 days: 1,510 visits, 64 bookings, 4.4% conversion, 58% show rate. Sales rated 81% of the enriched briefs as useful as or more useful than the old form data. Pipeline from demos roughly doubled. Total build time across the whole engagement: about 11 hours, most of it on the enrichment agent and its brief template.
The three-point show-rate dip is real, and we watch it weekly.
Thirty-three extra demos a month buys a lot of tolerance for a three-point dip.
One number we expected to move and did not: time-to-book. The shorter form shaved only eight seconds off median completion time. The win was never speed. It was dread.
What did not change
The skeptic's question is whether shorter forms just book worse leads. The 30-day data says mostly no. Qualified rate — sales' own tag after each call — went from 71% of demos to 66%, while absolute qualified demos went from 22 to 42 a month. The dip concentrates in very small companies that the old company-size field used to scare away. The client now routes those to a weekly group demo instead of a rep's calendar, which protects rep hours without turning anyone away.
Judge form changes on absolute qualified volume, never on rate alone. Rates flatter long forms.
The routing fix mattered here too. Deleting friction and then routing by segment kept rep calendars clean while volume doubled.
A form that filters by friction is not qualifying leads. It is losing them indiscriminately and calling the survivors qualified.
The rule we took from it
Every form field now has to pay rent: it must be essential to delivering the thing being requested, or demonstrably neutral to conversion, or replaceable by enrichment after submission. A demo booking needs a name, a work email, and a time slot. Almost everything else can be looked up, asked later, or dropped entirely. Run the field-level audit before any redesign — in our experience the fix is usually deletion, and deletion is free.
We have since run the same audit on four other client forms. Deletion won on three of them.
The client's sales team, initially the loudest objectors, now asks for the enrichment brief on inbound leads from every channel.
The fourth needed more fields — a budget question that saved both sides a wasted call. The audit decides. Not the aesthetic.