Taking sample size into account

when optimizing for ad placement

[Mister_Tut]

I know that improving one's CTR is largely a matter of tinkering with layout and optimizing ad placement. (There are, of course other factors like improving your content, but I digress...).

My problem is that I currently am getting few hits (average 58/day with an avarage daily CTR of 1.8%). With so few samples in my stats, will the results of any tinkering still be statistically valid?

If not, what would be a good threshold?

Would anyone care to recommend a time period for each trial based on stats similar to mine?

Thanks.

[hunderdown]

I was in a similar situation a year ago--similar # of clicks, though a better CTR.

I don't know if this is statistically valid or not, but as a rule of thumb I'd say that if you make a change, and your CTR immediately jumps up or down by a meaningful amount (say, up or down at least a third), and it stays near that level for three days, you're seeing a real change and not random variation.

For example, you're starting at 1.8% CTR. After a sitewide change, your CTR the next day is 3%. But the day after, it's 1%, and the day after that, it's 2.5%. That would tell me that the 3% might have been chance. On the other hand, if the three days after the change bring in 3%, 2.5%, and 3.5%, then you could conclude you were onto something.

Good luck!

[Mister_Tut]

Thanks. Three days might work if I was getting click-throughs every day, but my concern is that while my average daily CTR is 1.8%, I can go as long as four days with NO clicks, even with no site adjustments (beyond adding content- site is in blog format, but is a content site).

So I suppose this shows another defect in my understanding of what i should be doing. When I look at my CTR, over what period of time is it most useful to average it- quarterly, monthly or weekly? (only have 7 1/2 months of data).

Perhaps I should take your advice but multiply it by seven, so I'm looking at three weeks rather than three days.

The good news is that my total hits are continually increasing since I began learning SEO. Must keep plugging away... I'd really rather not wait 'till year's end for my first check!

[DamonHD]

Hi,

I think you should:

1) Change one thing at a time.

2) Measure your changed layout for at least a week for the normal weekly cycle to have passed.

3) Make sure that your site hits and other activity doesn't show unusual behaviour during this time.

4) Avoid major holidays and end-of-month as "weird" times!

I get a few more clicks than you, and generally don't follow the above rules strictly, but I do aspire to them. And I am now starting to automatically adjust/optimise ad placement. Too early to tell if it's working for me I think.

Rgds

Damon

[jetteroheller]

Use simple the formel

range=6.318*(Page Impression)-0.5047

This gives the plausibel change

CTR * (1+range)
CTR / (1+range)

An action to increase was successfull, when

new CTR / (1+range)

is greater than old CTR

When the CTR change is big, it can happen that 300 impressions are enough.

When the CTR change is small, more than 1000 impressions can be necessary

[moneyraker]

Going on 4 days with no clicks and yet getting an average of 58 clicks a day tells me that you can probably stratify your CTR data into at least 2 groups: days with almost zero clicks and days with large amounts of clicks. You must determine what's causing the two different levels of CTR (the day of the week?). You can then focus on improving the CTR of the poor-performing days by 'emulating' what happens on the 'good' days.

[Mister_Tut]

Well, to clarify, that's 58 HITS a day, not click-throughs. I've never had a day with zero hits. I've never had a day with more than 6 or 7 click-throughs.

Perversely, I just got slashdotted, so I can't apply any of these great suggestions for a little while now!

[hunderdown]

Oh, I thought that you had 58 clicks as a daily average....

At that level, honestly, I wouldn't worry about optimizing your ad placement. You need more traffic--that's a higher priority, I think.

Add content, get more incoming links, and repeat.

[Mister_Tut]

OK. Obviously I have been and continue to work on building my Hits Per Day. That was a really helpful comment- now I know where to focus my energy! More article submissions, more links, more content... Exactly the kind of insight I was looking for.

Thanks!