Home > Paid Traffic Sources > Push Traffic

---

How To Know When Statistical Significance Is Reached ? (6)

---

Hello!

I've coming across multiple affiliates talking about statistical significance I have a rough idea on the concept usually it's below than 5% but I'm quite lost on how to apply it to my campagnes, I would really appreciate some help!

---

Hi I came across this article and ss calculator - I hope this helps
article - https://charlesngo.com/statisticalsignificance/
calculator - https://www.houseofkaizen.com/conver...gnificance%2F=

---

Also use your common sense.

Statistical significance only works when you have conversions.

When you wait for statistical significance without getting enough conversions then it can happen that you overspend for a campaigns that´s not worth it.
It doesn´t make sense to run $0.20 payout offers for $150 or so just to wait for statistical significance.

When you don´t generate enough conversions to see that the campaign has potential you shouldn´t wait for statistical significance.
Then the campaign is just no winner.

First thing is to set a test budget and then check if the campaign is worth it or now.
In case of success you can use statistical significance calculator.

Don´t do it the other way round and wait first for statistical significance, no matter if the campaign is good or not.

---

This is pure gold! Thank you so much, What would you think a good place to start in terms of budget, currently trying PayoutX20.

---

Depends on the offer payout a lot. Let's say that you're trying something with $1 payout... in such case it's no problem to spend x20. But with a $25 payout offer, you probably wouldn't want to do that, especially with a lower budget.

Another thing to consider is whether you're about to test a new vertical, or you're just trying a new campaign within a vertical you have experience with and that you have proven creatives for. The more experience you have with something, the sooner you can tell that something has no chance of working.

And another thing is how many creatives/ads and LPs you're about to test. Because the more elements there are, the more data you need to buy.

This being said, 10x is like a bare minimum for a campaign with low amount of variations (ads,LPs...) and ideally, you'd want to give about 2x to every variable of the campaign. 20x for the whole campaign sounds more reasonable.

And just to set one thing straight, the 20x mentioned goes for like an initial test. I doesn't mean that after 20x you are either deeply negative or profitable. Let's just say that after such spend, you should be able to tell whether the campaign has any chance of working or whether the initial funnel shows some potential or you need to start from scratch again.

---

How to split-test landing pages or banner and cut inferior ones as they reach statistical significance:

https://stmforum.com/forum/showthrea...Banners-Part-1

How to do the same with offers:

https://stmforum.com/forum/showthrea...211#post289211

Hope that helps!



Amy

---

Home > Paid Traffic Sources > Push Traffic