In 2019, Dropbox made its largest acquisition to date: purchasing HelloSign for $230 million cash. HelloSign brought auxiliary document management tools like digital signing, online forms, and remote faxing into Dropbox’s growing collection of business offerings.
Prior to the acquisition, HelloFax’s web landing pages were functional, but dated.
One of HelloSign’s services was HelloFax, an online faxing solution that did not require a fax machine or landline. This is the page before Dropbox’s acquisition.
Dropbox quickly transitioned the branding on HelloSign’s webpages from HelloSign to Dropbox.
The same page after Dropbox’s acquisition. A quirkier style, more on-brand for Dropbox. It was called, “Faxing, Evolved.”
But after a few months, the conversion numbers on the Dropbox-branded pages were not matching their expectations.
Banner Year worked with Dropbox to create multivariate tests to measure the impact different design and copy choices had on conversion rates.
We designed, developed, and measured 2 new variations for each core service, exploring the response from users. We compared the conversion metrics of our variations with both HelloSign’s original design and Dropbox’s original redesign attempt.
Four versions of the same webpage. UPPER LEFT: the page as it was prior to Dropbox’s acquisition. UPPER RIGHT: Dropbox released a redesign post-acquisition called, “Faxing, Evolved.” LOWER LEFT: Banner Year’s first iteration called, “Tap. Faxed.” LOWER RIGHT: Banner Year’s second iteration called, “Jessica.”
At Banner Year, for clarity’s sake, we refer to each iteration of a design by a name that references something about its copy or imagery. For example, we called the Dropbox post-acquisition redesign, “Faxing, Evolved.” (This is better than referring to pages as “old,” “new,” “A,” or “B.”)
“Faxing, Evolved” looked more modern and hip at first glance. But it didn’t resonate with users in key ways that likely contributed to a lower conversion rate. We ran qualitative user testing sessions to understand what wasn’t working. Our key findings: