B2B software
Kestrel — 2.1× demo requests
Kestrel's home page tried to speak to four audiences at once. Visitors couldn't tell what the product did within the first screen, so most left before scrolling.
All case studies
The situation
Kestrel sells workflow software to mid-sized operations teams. The product is genuinely good and the sales team closed well once a conversation started — the problem was that conversations were not starting. Around 9,000 visitors a month arrived, and only 0.4% requested a demo.
The home page was the accumulated result of four years of internal compromise. Marketing wanted the enterprise story, sales wanted the SMB story, the founders wanted the platform story, and the solution had been a carousel that rotated all of them. The result said nothing to anybody, because a message that addresses four audiences addresses none of them.
Heatmaps showed 61% of visitors never scrolled past the first screen. Of those who did, the median time on page was 24 seconds. People were not rejecting the product; they were failing to work out what it was inside the window they were willing to give it.
What the audit found
Four messages competing
The rotating headline changed every five seconds. Reading research is consistent here: people skim, and a message that moves while being skimmed is not read at all. We measured a median of 1.4 rotations viewed — most people saw one slide, at random.
Internal language
The copy described the product using the vocabulary of the people who built it. Interviews with eleven recent customers produced a completely different set of words for the same thing, and none of those words appeared anywhere on the site.
A form as a qualification gate
Nine fields, including company size and budget range, before anyone could speak to a human. Qualification is a sales activity; doing it in a form filters out good prospects who simply do not want to disclose budget to a web page.
Proof without substance
A wall of customer logos with no figures attached. Logos establish that other companies bought it; they do not establish that buying it worked.
What was broken
- Four competing headlines rotated in a carousel nobody waited through.
- The product was described in internal language, not customer language.
- Demo requests sat behind a nine-field form asking for company size and budget.
- Proof was a wall of logos with no numbers attached.
What we changed
One sentence, one audience
The carousel was deleted and replaced with a single plain-language statement of what the product does and who it's for.
A form you can finish
Nine fields cut to three. Qualification moved to the follow-up call, where it belongs.
Proof with numbers
Logos replaced with three short customer results, each with a figure and a name attached.
The technical detail
The specific decisions behind the numbers, in plain language.
One sentence, one audience
The carousel was deleted. The first screen now carries a single plain statement of what the product does and who it is for, written in the words customers used in interviews, not the words the team used internally.
Three fields to a demo
Name, work email, and one free-text line describing the problem. Company size and budget are established on the call, by a person, where the answers are more honest anyway.
Proof with numbers attached
Three customer results, each with a figure, a named role and a sector. One quantified outcome from a comparable company outperforms twenty anonymous logos.
Separate pages per use case
Instead of one page trying to cover four audiences, each audience got its own page with its own language and its own evidence. Paid traffic now lands on the page matching the ad rather than a general home page.
Instrumented from the first screen
Scroll depth, section visibility and form field drop-off tracked per use case, so future copy changes are judged on data rather than opinion in a meeting.
How it ran
Research
Eleven customer interviews, win/loss review with sales, message testing against five alternative positioning statements.
Design
New home page and three use-case pages.
Build
Implementation, analytics instrumentation, split test setup.
Measure
Split test and pipeline tracking through the following quarter.
How we measured it
The new home page ran as a split test for four weeks across roughly 8,600 sessions, with demo request rate as the primary metric and qualified pipeline as the secondary one tracked over the following quarter.
Demo requests rose 2.1×, from 0.4% to 0.85% of sessions. Scroll-past-hero improved 34% and form drop-off fell 52%. Median time on page rose from 24 seconds to 1m 09s.
The number the business cared about most: sales-qualified opportunities per month rose 78% while the demo-to-opportunity ratio held steady, which confirmed the extra volume was not lower quality — the nine-field form had been filtering out good prospects, not bad ones.
demo requests
scroll past the hero
form drop-off
“Saying less felt risky. It doubled the pipeline in a quarter.”
What we took from it
- A carousel is usually an unresolved argument, published.
- Write with the words customers used in the last five sales calls.
- Every field in a demo form is a chance for a good prospect to change their mind.