Forum Moderators: martinibuster
How long does it take for earnings per click to go up because of the better conversion?
So you may have to wait a while. If you're getting the same number of clicks, you aren't losing out, so just sit tight and see if anything happens.
ecpm up 50% but no earnings diff
eCPM = Earnings / Impressions * 1000.
(SmartPricing reduces earnings before, not after, they appear in your stats).
SmartPricing reduces earnings before, not after, they appear in your stats
Yes, but if you make a change that affects the site-wide component of Smart Price, that doesn't take effect until up to a week later.
Carjocky, you've made an assumption that forums don't 'convert' well. We don't know that. Only Google knows the conversion rate of each part of your site, and they won't divulge for commercial reasons. Forums may typically have a low CTR, but those few clicks in your forum might have a high conversion rate for all we know.
To assess the impact of your change, monitor earnings for a couple of weeks to compare. If you want to work out whether smartprice has gone up or down as a result of this change, you'll need to refer to the post hunderdown alluded to:
[webmasterworld.com...]
Note, however, that I suspect there is also a page-related element of smartprice, the account-wide smartprice isn't the only component.
So, as I said before, wait and see. The AdSense + Smart Pricing algorithm is a black box. We can change our inputs and see what comes out at the other end, but may have no clue what's going on inside.
I think click-through rate (or something that correlates to it, but that we don't measure directly) MAY sometimes have an impact on smart pricing
In my stats, CTR and EPC have a zero correlation - this suggests there is no relation.
The main variables that correlate (negatively) with EPC are page impressions and clicks. That is probably a feature of demand and supply: more ad/referral supply depresses the price.
If I remove AdSense from pages generating 20% of my impressions but no or very few clicks, and if CTR for the site as a whole goes up as a result, what should happen to earnings? According to your theory, nothing much. But I've seen earnings go up by 30%, entirely due to an improved EPC.
So, like I said, CTR per se may not be factored into smart pricing, but on some site, some times, one of the things you can do to improve earnings is to remove the low-performing pages from AdSense. After you've tried and failed to get them to perform better, of course.
21_blue, that's your stats, not everyone's.
Agreed.
If I remove AdSense from pages generating 20% of my impressions but no or very few clicks, and if CTR for the site as a whole goes up as a result, what should happen to earnings? According to your theory, nothing much.
No, that doesn't necessarily follow. By remove Adsense from pages you are also reducing page impressions and clicks which, according to my stats, correlate negatively with EPC. So, "according to my theory", EPC may well go up when you do this.
You've actually changed three variables: impressions, clicks and CTR. And in your experiment they are all linked, so we can't necessarily tease out the cause/effect relationship, ie: which one is leading to the improved EPC, from this exercise.
There may also be other reasons for improved EPC in this experiment, nothing to do with SmartPrice. I'm just speculating but, for example, by reducing the number of exit points for your visitors you might be channelling them towards more expensive ads. For some sites this might result in a short term increase in earnings, but then a gradual decline if those redirected visitors convert less well as a result and hit site SmartPrice in the longer term.
What we can agree on is this: it's not a good idea to blindly slap AdSense code on every page on a site. One needs to analyze what's working and not working, and if a page or set of pages turns out to not work for AdSense, even after changing ad location/format/color, then do something else with it.
Impressions that don't generate clicks do you no good, unless of course if you've been targeted by a CPM advertiser. And if you have been, you should get ads back on every page on the site (or should you? What if the advertiser is looking at conversions and not just impressions...)
you should get ads back on every page on the site (or should you? What if the advertiser is looking at conversions and not just impressions...)
To some extent we are trying to untangle a plate of spaghetti whilst wearing blindfolds and gardening gloves.
But you may have something, here, because CPM Ads depressed earnings gradually on our site. This may have been related to conversions (the ads appeared on inappropriate pages). However, our "recovery" started immediately after opting out of CPM Ads - ie: we didn't have to wait until the next weekly update - so it wasn't due to solely to the site element of smartpricing!
Could you pass the parmesan, please.
Another thing we DO know about AdSense (maybe a new thread on this, "Things we know for sure about AdSense") is this: more good traffic (targeted, from relevant sites and searches) will lead to more earnings, other things being equal, though not perhaps in a linear way.
Yes, but if you make a change that affects the site-wide component of Smart Price, that doesn't take effect until up to a week later.
Again, SmartPricing can't play any role in this case (where eCPM went up while earnings remained the same). The simple equation that I posted shows that the only possible explanation is that the traffic and/or number of impressions went down. It's simple math.
eCPM = Earnings / Impressions * 1000
If eCPM goes up and Earnings remain the same, then Impressions must go down.
The issue is, do earnings go up? If earnings go up, then you've made something happen by taking those impressions off the table. You've either affected your user behavior, or you have affected smart pricing....
Ok so it has been close to a week and I still have at least doubled my CTR and eCPM BUT my earnings have still plumited.
carjocky, have you done the analysis I suggested on the first page of posts, to assess your smartprice changes? I did a similar experiment last year and found that earnings plummeted as well - I suspect this was because the low CTR ads we removed were also high converting ones. We put them back and earnings recovered.
Also, what really matters in my view is earnings per visitor. Has that increased/decreased/stayed the same?
Also, what really matters in my view is earnings per visitor. Has that increased/decreased/stayed the same?
That's a useful metric for e-commerce sites, but for publishers, eCPM is far more useful. Why? Let's look at a couple of reasons:
1) Unless you have a 100% reliable way to count visitors, your visitor numbers--and, as a result, your earnings-per-visitor numbers--are going to be estimates at best. (Plus, you have to define what you mean by "visitor" or "unique visitor," which opens a whole other can of worms.)
2) A site that uses AdSense may well have multiple sources of revenue (unlike, say, an Eddie Bauer or Best Buy e-commerce site), so what the owner really needs to know is how the AdSense ads are performing in comparison to other revenue streams. On my site, for example, I can quickly look at eCPMs for AdSense, display ads, affiliate program 1, affiliate program 2, etc. and see which source of revenue is pulling its weight and which is a low performer. This can help in determining whether a revenue source is worth keeping and which partners' ads or links merit better placement on the page.
for publishers, eCPM is far more useful...
1...visitor numbers--and, as a result, your earnings-per-visitor numbers--are going to be estimates at best.
2...what the owner really needs to know is how the AdSense ads are performing in comparison to other revenue streams.
EFV, as I've said recently (in another thread), I agree with you about point 2: eCPM is the best metric for comparing revenue streams. And although I didn't say it in the other thread, I also accept point 1: that visitor numbers are not firm numbers.
However, I disagree with your conclusion, based on this, that eCPM is therefore the best metric for the type of analysis being discussed in this thread.
Your point 2 is irrelevant because, in this case, carjocky isn't comparing revenue streams but trying to evaluate a particular site change.
And whilst the extent of impact of point 1 depends on how carjocky counts visitors, eCPM has far greater drawbacks for this particular type of analysis. If you ignore smartpricing for the moment, by definition if you remove ads from lower eCPM pages, then overall eCPM will increase and earnings drop. That is, there is an inverse relationship between eCPM and earnings when removing ads from pages.
If carjocky doesn't have visitor figures, or wants to use eCPM data because of the concerns you have raised, then a better CPM figure to use would be:
(page impressions for all pages, as reported in log files, not just those in Adsense reports)
To some degree, I don't think it matters too much which of these three numbers are used to divide earnings ("visitors", "unique visitors" or "page impressions from log files") as long as one doesn't use eCPM as reported by Adsense because, for this particular analysis, it is a totally misleading figure.
