Forum Moderators: martinibuster
I run a computer PC hardware content site (news, reviews, listings).
Anyone take a guess?
I think they had problems yesterday, when I got a BIG revenue (record)..
Now they are recuperating...
I have exactly the same type of site ( PC hardware content site with news, reviews, ...
I did not notice anything abnormal yesterday regarding the CPC.
Consitant mind I do have a broad spread over 700 clicks
CTR is holding steady as well, if not increasing a tiny bit.
Different industries different stats, too many variables to generalize across publishers.
Slightly off-topic and regarding the bottom line rather than CPC per se, I have a theory that every so often AdSense fires a barrel load of PSA's at a page or set of pages on a given site and then monitors the CTR to see if it drops or remains the same. If it doesn't drop, that puts up a flag that there may be fraudulent clicking taking place on those pages.
But it's just idle speculation. I might be completely wrong... >;->
I have a theory that every so often AdSense fires a barrel load of PSA's at a page or set of pages on a given site and then monitors the CTR to see if it drops or remains the same.
Interesting theory. It could be a worthwhile strategy, though I wouldn't see reason to do so for a site unless something else triggered it for investigation. I've been running AAs instead of PSAs since that became an option so I can't check, but do PSAs now have a tracking URL, because they never used to? I ask because I'm curious how you think Google could measure the PSA clicks. I know how they can for a subset of users, but I'm interested in your theory. Care to elaborate?
richmondsteve, it may not be the case that an initial suspicion triggers the PSA test. It may be something done as standard to all publishers.
There are a variety of ways it could be done. If it's done for hours or days at a stretch it would be noticed. If however one certain page always gets PSAs between 1.15 and 2.15 AM on Mondays, another page gets PSAs between 7.25 and 7.40 on Tuesdays Google could collate the data to come to some very useful conclusions on fraudulent clicks. And it's unlikely you or I would notice a pattern.
Google doesn't need to track which PSAs are being clicked. They only need to know which of your pages generated a click and whether they were running real ads or PSAs on that page at that time.
Or it may be the case that we are all just paranoid ;-)
in only 3 hours my EPC reached a normal level and is even higher with another 40-50%... So, a 100% grow in 3-4 hours
This only serves to demonstrate that the figures need to be seen in the wider context. Calculated over a period of a few hours they are completely worthless.
Google doesn't need to track which PSAs are being clicked. They only need to know which of your pages generated a click and whether they were running real ads or PSAs on that page at that time.
I agree, but I was really just asking ronin how he thought Google would track this. Last I checked PSAs didn't use tracking URLs (which is how clicks of Adword advertiser ads are tracked), just direct URLs. I know several ways PSA clicks could be tracked, but if the goal is to detect fraudulent behavior each has its limitations in terms of what would be measured. I was just hoping ronin could share more about his thoughts on that particular aspect of his theory.
What I posted before was the limit of my thoughts - and I might add that it was based on nothing but conjecture.
I'm not a programmer (apart from level 4 and a little of level 3, if I understand the terms correctly) so I have no idea how Google might use PSA click incidence to flag potentially fradulent clicks.
Nevertheless if I was responsible for delivering PPC ads, it's certainly something I would consider using as a safeguard against fradulent clicks.
I'm surprised to hear that there is no clicktrack code on PSAs (I've never looked at them that closely). If this is the case, the scenario I volunteered seems less likely, I agree.
When I said that they could be doing this now for a subset of clicks I meant anyone with the Google Toolbar installed in advanced mode since it sends data back to Google, including everything they'd need to track such data. But that would only detect fraudulent activity by such users and frankly I think most fraud is probably committed by bots, not users with the IE browser. They could also use the JavaScript onmousedown event trigger they currently use to detect an approximation of clicks for users with JavaScript enabled browsers in SERP results. If you don't know what I'm referring to view the source of a Google SERP and look for the "clk" JS function.
Anyway, attempting to detect fraud using the method you described is a pretty interesting concept. And someday we might find out they're doing it.
I've already contacted AdSonar to start lining up a replacement for AdSense. The downward trend with Adsense earnings is continuing, even appears to be accelerating.
Ignore the spikes and watch the curves: Long-term trends are what matter.
Fewer impressions, fewer clicks and quadruple the revenue for today compared to any day this past week.
I think it has a lot to do with the way the SERPs have been moving lately.
Long-term trends are what matter.
I agree with Europe. And long-term is not a couple of weeks. Until you have been on the program for at least a year, you aren't going to be able to tell what your true trends are. You need at least a year to be able to factor in phenomenon like seasonable adjustments that may impact your market niche. Consumers simply don't act the same in from month-to-month or season-to-season. Advertisers are the same. They don't budget advertising dollars in a mechanical flat line. It is a dynamic process.
How do you know what your January EPC, CPM, income, etc. for AdSense should be? Have you ever experienced a January yet with AdSense?
This fixation on extreme short-term fluctuation is simply a time wasting exercise. Post again when you can tell me you're off 20% from your seasonally adjusted historical average for the time period. That would be useful.
