Forum Moderators: martinibuster
Do they consider multiple clicks from one IP or cookie one impression?
Our page counts are not corresponding to their impressions at all for the past few weeks. GAS is running at about only 25% of the total we get from the web server reports. So this leads me to believe that they are looking at IPs and counting only one view per. OF course that does mean the CTR will go up if a lot of people click on the ads, and we are seeing a rise there, but I would like to make sure all of our page views are in fact being counted.
BTW, I was on the cbs.com site a few days ago and their GAS ads were coming up completely blank ... just white space for several refreshes. Not even showing PSA ads. wild.
Thanks
It could be due to your site being hit with many pageviews by web spiders, or due to counting pageviews that do not have AdSense ads on them, or even to the reporting of "hits" (which includes images and other stuff) as opposed to true page views.
Does anyone know for sure what GAS (that's an interesting acronym, eh?) is using to calculate IMPRESSIONS?
Yes. See AdSense FAQ 2. What is a page impression? [google.com] Based on analysis of my own logs differences in Google reported impressions and log or stats package reported page views is entirely attributal to what's stated in the FAQ. Also, keep in mind that Google's reporting day is based on the PST timezone.
Do they consider multiple clicks from one IP or cookie one impression?
I'll assume that you mean "views", not "clicks". Multiple impressions from the same user are recorded as such. My reported impressions are always several times larger than the total # of IP addresses reported for human visitors.
GAS is running at about only 25% of the total we get from the web server reports.
You might be surprised at how many of the page views recorded in your web logs or reported by your reporting tools are in fact bots (non-human visitors) that don't load the AdSense JavaScript code. Also, page views by humans that have JavaScript disabled or don't allow the page to fully load so that AdSense loads will not be reported as impressions by AdSense. Also reporting tools sometimes report images, and other files that aren't actually web pages as page views as well as pages that are returned with something other than a 200 status code. When I remove bot page views from my logs my page views are always within 5% of the totals reported by AdSense so I feel that they're accurate.
We run a bunch of servers here and are quite familiar with how web server logs work. Our reporting tools only look at how many time a page loads, never counting any of the other links or images on the page, etc. - they get filtered out of the reports. We just check the root page loads for the core page. And we know where the clicks came from .... this particular day the increase of 5 times normal was a special event, not from spiders or bots. We can get the reference incoming links for all pageviews from the reports.
I can also get a visitor count report, which is more in line with what the count is from GAS. A visitor count only looks at how many times a request comes in from an IP address. Problem is that AOL and other caching services all look like the same IP at times even though we have the META expire on the pages down to seconds.
I find it hard to believe that only in the past few weeks people stopped using JavaScript or stopped letting pages load. We haven't changed anything here. Have THAT many people installed Norton firewall in the past month and are using the default NO ADS setting?
We had a major balloon of traffic in the past few days that really pointed out the count issues. We were getting hundreds of page views per minute at one point. This particular page ended up doing like 5 times what it normally does and it will go back to normal later today. It's an event based page and the event is over for now.
That's why we would REALLY like to see some kind of accounting by page so we can get a handle on which pages are producing the most traffic and when. I can live with knowing that a page didn't produce click-throughs becuz of the normal reasons we have all discussed here before, but I have trouble now knowing what the page count discrepancies are.
That's why we would REALLY like to see some kind of accounting by page so we can get a handle on which pages are producing the most traffic and when.
Doesn't Webalizer do that?
I had been shrugging it off as maybe real, maybe people just overreacting to short-term fluctuations. However Google are definitely undercounting impressions on my site today, by about ~50% (clicks and earnings look right). There's another thread about it here:
[webmasterworld.com ]
I forgot to include the fact that a few days ago we got about 80% of the activity we did today and it looks like ALL the page views were counted since the web anlayzer reports are pretty close to what GAS reported. That day's site activity was 80% of today's, yet today's activity on GAS shows less than 50% of what the other day showed. It should show 20% more than the other day, so it's showing about 70% less then it should. So today's count isn't even close to what it should be and I don't see why they would all of a sudden catch up. I know some of you have the opinion that you need to wait and see, but unless you can track these things in real time, it is awful hard and much more time consuming to go back and see what happened after a few weeks and in the meantime you were operating with something that possibly needed fixing.
