Removed low performing channels, now what?

When does income pick back up to match higher CTRs?

[bostongio]

Based upon all the recent discussion regarding Smart Pricing and lower-performing sites/channels, I removed adsense from all of my very, very low performing channels. Of course, overall traffic numbers went down (one of these channels were some forums, very low CTR on them) and my overall income has also gone down. It's been about 10 days now.

I see my CTR starting to climb (almost double) after the first week. However income isn't following that trend -- it's basically gone down a little (not significantly yet).

After I make a change of this proportion, how long should I wait to decide whether it's "worth it" for me? I mean, it's great to double my CTRs and all, but that's not yet translating into income. Should I wait a month? 2 months?

Thanks!

[21_blue]

Hunderdown wrote:
>I'd suggest trying it for a full week or longer--since Smart Pricing doesn't update daily.

See [webmasterworld.com...] for a method of working out when Smart Pricing updates on your account.

[hunderdown]

21_blue, this is where the whole thing gets too complicated for me to deal with!

Were those pages also low EPC? I honestly don't remember, and I guess I should go back and see.

But it's a good question, because I see enormous variation in click value on my site.

As a side note, my current problem is that CTR has dropped by about a third recently. I can't figure out if it's seasonal, related to my site reorganization, due to expanded ads, or all three.

If Smart Pricing were all we had to contend with, then life would be simple, wouldn't it?

[martingale]

hunder:

It could still be "site-wide page CTR" that matters. You knocking off a poorly performing page would up the average page-CTR for the whole site. My knocking off the footer ad block on a page with 3 adsense blocks might not have the same effect.

Anyway, we will see. For now I have replaced the adsense footer with an amazon recommends ad block.

However.. I will probably learn nothing... my revenue has been ramping up over the past few weeks as we approach the end of the year anyway (my site is seasonal, with the next couple of months being my best)... that effect will confound any attempt to work out whether it's a CTR thing.

[Thez]

I'm still at the same level after removing the poor channels, the ecpm has stayed up ever since.

[Nitrous]

I wouldnt get too excited, I changed nothing and fri, sat, sun all 30 percent better than usual £, with the same 400 clicks...

Weekends are normally down 15 to 20 percent but not this one!

[Jobarr]

I'd like to try this - around what CTR should I use as a threshold for which ads to remove? 1%? Lower, higher?

[janethuggard]

"All the advice to the effect that less ad blocks can mean moe income is just plain wrong. You'll just earn less and less by removing ads from "low performing channels"

You need to do some testing. Our changes resulted in double the income overall on the test sites. The modifications were a huge success, and are now being implemented network wide.

[hunderdown]

Jobarr, the number is going to vary. Assuming that you've got your pages broken out into a good number of channels, one approach is to simply remove the ads from the worst-performing one (look at both CTR and earnings--I'd be hesitant to kill a channel with a 0.5% CTR but the occasional $5 click, for example). See what happens. Give it a week. Then try the next worst one, and so on.

Or look at your site average, and remove ads from all the pages with a CTR less than one tenth of it. Or some other fraction.

Don't overdo it. If it doesn't increase overall earnings, you may have done too much. If it hurts overall earnings, you have definitely done too much.

And, of course, on channels that are marginal--some clicks but not great--try to improve the CTR first.

[Thez]

"The 70% increase, in itself, is not particularly meaningful because it depends on what the "standard deviation" is of your site. If your standard deviation is, in effect, 10% of your average earnings, then your proof is very compelling; if it is 100% of average earnings then the 70% change could simply be a coincidence."

I got my hands on an excel and calculated the deviation.

Two weeks before the change, deviation was 14,2. Since the change (10. nov - to date) deviation has been 15,4. Averages 104,3 and 148,2.

Hope that helps. :)

[21_blue]

Thez wrote:
>Hope that helps. :)

Yes, it does. But it is worth noting what it actually suggests: that your 30% average increase in earnings is statistically significant - because it appears to be of the order of a couple of standard deviations. What it doesn't do, necessarily, is attribute causality. That is, is your increase in earnings due to the changes you made, or is it due to other factors?

Most of our changes in earnings are due to changes in visitor numbers - earnings per visitor generally vary within a range almost irrespective of the tweaks we make to the site. However, I'm still waiting for the full SmartPrice impact of recent changes I made to feed through this Thursday/Friday, so we'll see if earnings per visitor go up then or not.