Case StudiesCase study
DisplaySplit
DisplaySplit is a Windows app for people who replaced two monitors with one ultrawide. It makes Windows see the panel as separate displays, so sharing a screen, going fullscreen and the taskbar work the way they did with two monitors. I designed the product, its brand, the app and the site. It is in pre-launch, with the site live and a waiting list open.
The problem
An ultrawide replaces two monitors on the desk, but not in Windows. Share it in a call and colleagues see everything at 37.5% of its size.
What I did
Defined a product that makes Windows count one panel as two displays, then designed its brand, the app and the site, with the most care on the one risky moment: the switch.
What it came to
- 2Real displays from one panel
- 5Steps to set up
- 15sTo undo any change
01 / The problem
An ultrawide replaces two monitors on the desk. Not in Windows.
People buy a 49-inch ultrawide to replace two monitors: one cable, and no bezel down the middle. The panel has exactly the pixels of two 2560 × 1440 screens side by side. Windows counts it as one display, and so does every program.
That shows up in the same four moments of an ordinary working day. The worst is a call. Share an ultrawide and a colleague on a laptop sees the whole strip squeezed to 37.5% of its size, with black bars above and below.
Checking it isn't only me: requests to Microsoft since 2020
The request already had a name: virtual monitors.
Teams community, 2021
"Illegible for others"A thread on sharing 42 to 49-inch screens, 32 replies long. The answer given: Teams shares a window or the whole desktop, and nothing in between.PowerToys, 2020
"Create virtual monitors"A request to Microsoft's own window-zoning tool, so a maximized app fills a zone and not the panel. Closed as a duplicate.PowerToys, 2026
Still openThe same request, filed again in May 2026: fullscreen that respects the boundary, a taskbar for each side, and sharing one side in a call. It is the four moments above, written by someone else.Sources: Microsoft Tech Community, "Sharing part of the screen"; PowerToys issues 6080 and 47571.
What people do today, and where each fix stops.
A window manager sorts windows into zones, but the programs still see one display, so calls, fullscreen and the taskbar don't change. The monitor's own split mode does give Windows two displays, but it takes a second cable and a trip through the monitor's menu, and on a monitor like mine it lowers the refresh rate. Going back to two monitors undoes the reason for buying one.
So the job was specific: make Windows count one panel as two monitors, on one cable, from the desktop, and make it safe to switch.
02 / My role
Self-initiated, from the idea to pre-launch.
I own the product and every decision on this page.
Mine
- Product definition: what it does, what it refuses to do, how it is sold
- The brand and the visual system
- The app: overview, layouts, setup and settings
- The switch, and the safety model around it
- The marketing site and its sign-up
What I made
- A Windows app: six screens and a five-step setup
- One design system for the app and the site
- A nine-section site with a working demo
03 / Goals
What solved would mean.
Three things. The results further down are set against them.
Goal 1
It behaves like two monitorsSharing, fullscreen, maximize and the taskbar all work per side, with nothing to set up per app.Goal 2
It is safe to switchNo display change can leave someone unable to see it or undo it.Goal 3
It is ready in a minuteFrom first launch to a working split without reading anything, and back to one panel in one action.
04 / Decisions
Five decisions, and what they cost.
1. A real split, not an arrangement.
The obvious product arranges windows, and my first version did. It worked, and it failed the first goal, because programs still saw one display. I threw it out and chose the harder route: each side is a display Windows itself recognises, so every program gets the right answer without knowing DisplaySplit exists.
- Gave up
- A simple install. A display driver needs administrator approval and has more ways to fail.
- Got
- Every program behaves correctly, with nothing to set up per app.
2. State in a sentence, one switch.
The app opens on a sentence, "Split is off", with one line under it saying what that means and one button. Below that, the panel is drawn to scale beside what it will become. Layouts are seven presets with plain names, or a divider you drag, and each screen shows the size it will be before anything changes.
- Gave up
- A dashboard. There are no graphs, tiles or tabs on the first screen.
- Got
- Anyone can tell what state their display is in, and how to change it, at a glance.
3. Setup that changes nothing until the end.
The wizard checks the display, lets you choose a layout and asks for a licence, in five steps with one decision each. Nothing on screen changes until the last step, and that step names both safety nets right next to the button.
- Gave up
- A one-click start.
- Got
- Nobody's display changes before they have seen what it will become.
4. The switch gets the most design.
A display change can fail in a way you can't click out of, so each part of that moment has its own piece of interface. Before: the switch isn't offered until the checks pass. During: Windows takes a few seconds to rearrange everything, so a quiet full-screen cover says "Splitting your display" and lifts when it has settled. After: a box asks "Keep this layout?" and counts down from 15, and doing nothing puts everything back.
And always: one shortcut, Ctrl + Alt + Shift + F12, returns to a single display from anywhere.
- Gave up
- Speed. Every layout change can cost fifteen seconds of confirmation.
- Got
- Nobody is left looking at a screen they can't use.
5. A second pass on the interface.
The first pass worked, and it could have belonged to any app: a grey panel, a standard blue button, one rectangle per screen in two unrelated colours. Nothing in it said what the product was, and this is a paid product with no free tier. I set a visual direction and rebuilt every screen to it: darker, quieter, one accent, with the panel itself as the picture.
- Gave up
- A finished interface that already worked.
- Got
- One that looks like a product you would pay for.
05 / The system
One blue, light type.
Ink for the surface, one blue for anything that acts, and paper for the daylight half of the site. Headlines are light, large and set in two tones, so the second line carries the point. Labels are small mono capitals.
The display tile, white running to blue, is the one picture the product owns. It is the app's preview, the wizard's layout control and the site's hero. The site is set in Host Grotesk and Geist Mono. The app uses Segoe UI Variable, the typeface Windows already ships, at the same light weight, so it feels native and is still recognisably the same product.
06 / The site
Every picture is the product.
There is no stock art on the page. The hero is the panel itself, coming apart as you pick a layout. Four small drawings in the same language show what changes. A demo lets you switch the split on and watch fullscreen, sharing and maximize change.
One section is called "What it doesn't do", under the line "Better to know now than to find out after paying." A product that asks for money before it works for you has to be straight about its limits first.
07 / Results
Against the three goals.
On a 49-inch 5120 × 1440 panel, with the app running:
Goal 1 · Two monitors
Windows lists two displaysEach has its own taskbar and a maximized window fills one of them. Every program sees two screens, with nothing set up per app.Goal 2 · Safe to switch
Four layers, no dead endsA check before, a cover during, a question after and a shortcut always. Doing nothing always ends on a working display.Goal 3 · A minute
Five steps, one switchOne decision per step to set up, one button on the first screen after that, and one shortcut back to a single panel.
The site is live at displaysplit.com with the waiting list open.
What I'll measure at launch.
- Sign-up to download to licence. The app itself reports nothing, by design, so this is read from the site and the checkout.
- Support mail that mentions a stuck or black screen. That is the real test of the safety model.
- Refund requests, and the reason given. The limits page is there to keep that number small.
What I'd take to a team.
I started with the wrong product and only saw it by writing down what solved would mean. And the engine is the part nobody sees. The product is what happens before, during and after the one moment that can go wrong.
*The app screens on this page are the real app, running with a test licence. The cover, the confirm box and the checks in the switch picture are redrawn from it.