Statistical Significance for Impressions

[Freejoy]

How many impressions before the CTR becomes 95% accurate?

 

[Graybeard]

hypothetically ...
500 clicks (CTR is statistically valid [1,000+ is better])
so what is the impressions CTR and where is it? Banners on mediocre websites or PPC ads in Google?
100 impressions in Google Ads (AdWords new name) may produce 3-7 or more click-throughs.

banner impressions 50000*0.00015 =7.50000
340000*.00150=510.00000

It might take 340K banners 510 CTR and 10 - 50 payable events (worst case scenario)

 

[awesomesauce]

^^ what math do you use?

If you use a Bayesian calculator it will depend on how many clicks you get on those impressions. Comparing 2 ads... 500 impressions on each ad... if one ad gets 1 click and the other 5 clicks.... it will be 95% accurate the one with 5 clicks is better.

 

[Freejoy]

Thank you.

So after getting about 500 - 1000 impressions the CTR should be about 95% accurate?

 

[awesomesauce]

Thank you.

So after getting about 500 - 1000 impressions the CTR should be about 95% accurate?

As I said, it depends on how many clicks you get on those impressions.

 

[Graybeard]

Knock yourself out
NO, clicks ON impressions
PPC or some other display and what quality is the media?

You cannot compare ALL impressions from any source ...

Banners or pop ads you may need a A LOT of CPM to get 500 - 1000 clicks.
Unless, bots and network unfiltered garbage 'counts'.
Garbage traffic makes real stats NEAR impossible.

Bottom line is always customer acquisition cost per media/campaign.
That is the ONLY stat that matters when you go to the bank

  <SCRIPT language="LiveScript">
<!-- hide this script tag's contents from old browsers

var p1; var p2; var p3; var p4; var p5
var r1; var r2; var r3; var r4; var r5
var p

function CalcStats(form) {

r1=0; r2=0; r3=0; r4=0; r5=0

p1=eval(form.pp1.value)
p2=eval(form.pp2.value)
p3=eval(form.pp3.value)
p4=eval(form.pp4.value)
p5=eval(form.pp5.value)

if(form.o1.checked=="1") {
r1=r1+p1*eval(form.a11.value)
r2=r2+p2*eval(form.a12.value)
r3=r3+p3*eval(form.a13.value)
r4=r4+p4*eval(form.a14.value)
r5=r5+p5*eval(form.a15.value)
}

if(form.o2.checked=="1") {
r1=r1+p1*eval(form.a21.value)
r2=r2+p2*eval(form.a22.value)
r3=r3+p3*eval(form.a23.value)
r4=r4+p4*eval(form.a24.value)
r5=r5+p5*eval(form.a25.value)
}

if(form.o3.checked=="1") {
r1=r1+p1*eval(form.a31.value)
r2=r2+p2*eval(form.a32.value)
r3=r3+p3*eval(form.a33.value)
r4=r4+p4*eval(form.a34.value)
r5=r5+p5*eval(form.a35.value)
}

if(form.o4.checked=="1") {
r1=r1+p1*eval(form.a41.value)
r2=r2+p2*eval(form.a42.value)
r3=r3+p3*eval(form.a43.value)
r4=r4+p4*eval(form.a44.value)
r5=r5+p5*eval(form.a45.value)
}

if(form.o5.checked=="1") {
r1=r1+p1*eval(form.a51.value)
r2=r2+p2*eval(form.a52.value)
r3=r3+p3*eval(form.a53.value)
r4=r4+p4*eval(form.a54.value)
r5=r5+p5*eval(form.a55.value)
}

p=r1+r2+r3+r4+r5
if(p==0) {p=1}
r1=r1/p;  r2=r2/p; r3=r3/p; r4=r4/p; r5=r5/p

form.rp1.value = Fmt(r1)
form.rp2.value = Fmt(r2)
form.rp3.value = Fmt(r3)
form.rp4.value = Fmt(r4)
form.rp5.value = Fmt(r5)

}

function Fmt(x) {
var v
if(x>=0) { v=''+(x+0.0005) } else { v=''+(x-0.0005) }
return v.substring(0,v.indexOf('.')+4)
}

function RevPrior(form) {
    form.pp1.value = form.rp1.value
    form.pp2.value = form.rp2.value
    form.pp3.value = form.rp3.value
    form.pp4.value = form.rp4.value
    form.pp5.value = form.rp5.value
    }

<!-- done hiding from old browsers -->
</SCRIPT>