Why running them at the same time matters
Sequential tests compare two time periods as much as two pages. A quiet fortnight, a bank holiday, a mention somewhere that sent unusual traffic, and the result is about the weeks rather than the pages.
Rotation removes that. Both destinations get the same days, the same mix of channels and the same audience. Whatever difference remains is more likely to be about the pages.
This is not a formal experiment and it is not going to satisfy a statistician. It is a great deal better than guessing, and it costs one setting on a link you were going to make anyway.
Setting one up
- Create the short link and add both destinations to the rotation.
- Give each destination its own UTM content value, so your own analytics can tell them apart after the click.
- Share the single short address everywhere. Nobody outside needs to know a test is running.
- Leave it alone. Changing anything mid test means starting again.
The last step is the one people skip. A test you keep adjusting is not a test, it is a series of very short tests with no useful data in any of them.
Change one thing
If version B has a different headline, a different image, a different price and a shorter form, and it wins, you have learned that four changes together beat none of them. You have not learned which one to keep.
Test the thing you would actually change. Usually that is the headline or the first screen, because that is where people decide whether to stay.
The second most useful test is the destination itself. Sending the same audience to a landing page rather than a homepage often produces a larger difference than any wording change on either.
How long to run it
Longer than feels necessary. Early numbers swing wildly and the version that leads in the first fifty clicks is often not the one that leads at five hundred.
Run at least one full week so every day of the week is represented, and wait for a few hundred clicks per version before drawing a conclusion. If your traffic will not produce that in a reasonable time, the difference you are looking for is probably too small to matter anyway.
Ignore the first day entirely. It is mostly you, your colleagues and whoever you sent the link to for testing.
Reading the result honestly
Clicks tell you how many people arrived at each destination, which is only half the question. The other half is what they did next, and that lives in your own analytics rather than in the link.
A version that gets fewer clicks but more completed bookings is the better version. Judge on the outcome you actually want, not the one that is easiest to see.
And accept the boring result when it comes. Most tests end with no meaningful difference, which is genuinely useful information: it means you can stop arguing about that thing and go and change something that matters.
After the test
Point the link at the winner and keep the same address. Anything already shared moves to the winning version without anyone doing anything.
Write down what you tested and what happened, even when nothing happened. Six months later somebody will propose the same test again, and a one line note saves a fortnight.
Then pick the next thing. The value is in the habit rather than in any single result.