How the test works
Your page against the one you run now, on your own ads, with the win agreed in writing before anybody sees either page.
Last updated 23 August 2026.
The one number we compare
Out of every hundred people who click your ad, how many end up buying. That is the number, counted the same way on both pages, from the click through to the order.
Beside it we read what each click was worth in money. That way a page that sells more of the cheap thing cannot pass as the winner.
How many people move from the pre-sell page through to your store is a number we watch, and it never decides anything. The end is the part that pays you.
How the split works
Your ads stay exactly as they are. Same campaign, same ads, same budget, same days. The only thing that changes is where the click lands.
Some of the clicks go to the page you run now and the rest go to the page we built. The share is agreed before we start, and nobody picks who sees which one. It happens at random as each person arrives.
Your ads can keep rotating and keep getting edited while the test runs. Whatever happens to them happens to both pages at once.
What we set up before any money is spent
The counting has to be right before anything runs, so we prove it first.
We check that a click keeps its tag the whole way from the ad to the order. No order can then land against the wrong page. Your own analytics gets checked too. Left alone, it reads the step from our page to your store as a fresh visit from somewhere else. And your pixel goes on the page we built, so the ad platform sees what your store sees.
Then two real orders go through. Each one is followed end to end, until it lands with the right page against it. Only then does a single ad point anywhere new.
Some of this needs a hand from whoever runs your ads. Anything we need is written down before the test opens.
What gets written down before it starts
Before either page has a visitor, we both agree one page of plain words.
What counts as beating it gets written down and agreed before the test starts, so neither of us gets to decide afterwards what the word meant.
That page also says where you are starting from, taken from your own numbers and not from our guess. It names the ads that will feed the test and the share of traffic each page gets. The last two lines are the orders that have to land before anyone calls it, and how long the test runs.
How long it runs, and when we call it
It runs until the orders we agreed on have landed. Not until it looks good, and not until somebody wants an answer.
Ask at any point and you get the raw numbers as they stand that day. A winner is the one thing we hold back until the count is there. A small pile of orders points one way this week and the other way the next.
If our page loses
We build you another one, with a different angle, free. One rebuild per test, and it takes the production slot your next page would have had. There is no money back, on any plan, and we would rather say that here than let you find it later.
What makes a test not count
Whatever changes has to change for both pages, or the reading gets thrown away.
Swap the ads halfway through and the two pages are no longer being sold the same way. The number stops meaning anything. Same for a price that moves on one side only, or a discount that runs to one page and not the other.
The rule points at us as well. If we edit our own page in the middle of a test, that reading is dead. We say so, and the count starts again.
What we never touch
Your product page stays exactly as it is. What we build is a separate page at its own web address on your own domain. We hold no login to your store.
You add one line to your own domain settings, once, and it changes nothing that is already there. We never hold that login either.
Your ads stay yours. A fair test swaps the page and nothing else.
If something here is not what you expected
Tell us before the test starts rather than after. Write to bas@preselly.com and a real person reads it. What the offer costs and what you own at the end is on the terms page.