01The business
A manufacturer of in-wall mounts for tablets and touch screens. It sells through dealers, and installers who want to become dealers fill in an application form on the website. Every submission is a potential reseller. The business is described by type on this page, because naming it needs the owner’s permission.
02What WeaverAll noticed
WeaverAll counted the people who opened the application page and then submitted the form, by the device they used, from 27 September to 20 October 2026.
- Computer: 40 of 131, 31 in 100.
- iPhone: 13 of 86, 15 in 100.
- Browser inside the Facebook app: 5 of 59, 8 in 100.
- Browser inside the Instagram app: 4 of 52, 8 in 100.
- Android: 2 of 96, 2 in 100.
All phones together: 25 of 295, which is 8.5 in 100. Phones are seven in ten of everyone who opens the page. The gap holds inside the same source: visitors from unpaid Google search submitted 34 in 100 times on a computer and 8 in 100 on a phone. Visitors who typed the address did 28 and 11.
The form itself asks for about two dozen items, twelve of them required, including a tax number, the number of employees, a mailing address and a contact for homeowner enquiries. Country is typed by hand; in these weeks the United States was typed six different ways.
03The suggested change
Two steps, in order. First, open the form on an Android phone and check that every field and the submit button work, because 2 in 100 on Android against 15 in 100 on iPhone points to something broken rather than something long.
Then test a short phone version: company, name, email, phone and country, with the rest requested later by email. Or keep one form and make the tax number, the employee count and the homeowner contact optional.
The gap shows inside the same source: visitors from unpaid Google search submit at 34 in 100 on a computer and 8 in 100 on a phone.
04How success is measured
The share of phone visitors who submit the form, which is 8.5 in 100 today, with the computer rate of 31 in 100 as the comparison group. The site cannot split visitors at random, so the computer rate stands in for “what would have happened anyway”: if both rates rise together, something else changed. First check after four weeks, which is about 350 phone visitors.
The report also put a size on it: if the phone rate rose from 9 to 15 in 100, the 295 phone visitors of those 24 days would have produced 44 applications instead of 25. That is an estimate to test, not a promise.
What it could be worth. About 19 more applications in 24 days is about 24 a month. WeaverAll gives no money figure, because a dealer application has no price on the site and only the owner knows what a new dealer is worth. The owner can multiply: 24 applications a month, times the share that become active dealers, times what a dealer orders in a year.
05Approval and the change
The improvement report is dated 21 October 2026. As of 10 November, WeaverAll has no record of a decision. The Loop is waiting.
06What happened
Nothing to measure yet. This page will be updated when the Android check is done and when four weeks after a change are in.
07What we learned
Count by device before you redesign anything. A single completion rate for the page, 15 in 100, hides a computer rate twice as high and an Android rate near zero.
Check for breakage before you shorten. A long form explains 15 in 100 on iPhone. It does not explain 2 in 100 on Android.
08Honest notes
The data shows who submitted, not why the others stopped. Form length, a fault on Android and a different kind of visitor on phones can each fit these numbers, and the Android check is there to tell them apart. The site’s other form, “find a dealer”, has six required fields and a much smaller gap: 13 in 100 on a computer and 10 in 100 on a phone. Leaving out the 55 phone visitors who came from a video ad moves the phone rate from 8.5 to 10 in 100. Five test applications by the owner’s team are left out of every figure. The business is not named because naming it needs the owner’s permission.
Real story. The business is a customer of WeaverAll and is described by type until its owner agrees to be named. All counts come from WeaverAll's records of the site's own traffic for 27 September to 20 October 2026 and from the improvement report dated 21 October 2026. Nothing on this page is illustrative.