Mechanism
One paid click has to do two jobs: send a real person to the offer, and leave behind an audience worth advertising to later. This is how the two pages split that work, and what to watch once traffic starts.
A large share of pop and popunder traffic never had a person behind it. Those clicks still reach the offer, still cost money, and still dilute every number you look at afterwards.
You cannot filter that with a checkbox, because a script clicks checkboxes. What a script does not do reliably is complete a continuous gesture and release it in the right place. So the exit is priced in effort rather than in clicks.
The second half of the problem is waste of a different kind. Even a real visitor who does not install is worth something, provided you can reach them again. A single page throws that away the moment they leave.
One entry, one gesture, two destinations opened at the same moment.
Both destinations come from the same release, in a fixed order. The visitor ends up looking at the offer, with the second page waiting behind it in the tab they started from.
The gate is not there to annoy anyone. It is there to make the exit cost something an automated visit will not spend.
A partial slide is not a rejection. The pad eases back and the panel stays open, so a person who fumbles it simply does it again. Nothing is spent and nothing is lost.
The same click pays for both. Each page spends it on a different thing.
It has one job and one exit. Creative, a short promise, and the gate. Nothing on it competes with the button, and there is no second link to leak attention.
It loads in the original tab and is the place the tracking pixel fires, before any interaction, so even a visitor who reads nothing and leaves joins an audience you can advertise to later. No pixel is installed at the moment.
It does not ask for the gesture again. By the time someone lands there they have already proven intent once, so the button is a plain link. The page exists to hold the tab, keep the offer within reach, and record the visit.
The offer only pays on Windows desktop. Every click from a phone is spend with no possible return.
Both pages check the operating system and device type in the browser before anything else. On a phone, a tablet, a Mac or a Linux machine, the button changes to a notice and a copy-link field so the visitor can move to a computer. The exit is never armed on those devices, so there is no path to the offer at all.
Mobile exclusion belongs in the targeting of the traffic source. The gate exists to catch the spend that slips through anyway, not to replace the campaign setup.
Same funnel, same copy, same gate. Only the creative and the arrangement change, so any difference in the numbers comes from the creative.
| Version | Creative | What it is testing |
|---|---|---|
| 1 | Still image, animated in the browser | The cheapest and fastest to paint. The baseline everything else has to beat, and the fallback if load time turns out to matter more than movement. |
| 3 | Two characters, video loop | Whether attention can be steered. The button sits in the gap between them so both lines of sight point at it, and it reacts each time the loop returns to that moment. |
| 4 | Vehicles, video loop | Whether speed and machines sell the open world better than characters do. Also a different reading order, with the copy beside the button rather than above it. |
Run them against the same source and the same targeting, or the comparison means nothing.