Are your Google Page Impression Stats correct?

[yobo]

Google said I served up 3000 page impressions yesterday. My stats say 10,000 page views. Anybody else seeing a huge discrepency? Curious?

[JohnKelly]

Google on counts impressions where both Javascript and Iframes are loaded. Your stats likely include search engine spiders and all other types of page loads.

[ByronM]

anyone know of the PSA's are counted as well?

[Jenstar]

Yes, PSAs are included. I have noticed the stats being behind - they might catch up by the end of the day as well. Are the clicks and earnings about what they should be?

Also, were you one of the ones showing the "data not found" problem a couple of days ago? I wonder if it could be related, since at that time some showed regular stats, while others had that error.

[jomaxx]

yobo, robots of various kinds do not tend to trigger AdSense impressions but can be responsible for many many pageviews. Is 10,000 a typical number of impressions for your site, or was yesterday a spike?

[yobo]

It's typical. Its definitly not robots.

[UKFord]

Same here. Google shows numbers less than 20 per cent of what the server stats are.

[mcavill]

>>Google shows numbers less than 20 per cent of what the server stats are.

That could (almost) be due to browsers with JavaScript disabled.

[danny]

Robots and spiders are a big fraction of my page fetches. Googlebot alone pulls every pretty much every one of my page every day! Throw in the ordinary users who are using some kind of automated prefetching - which probably doesn't evaluate Javascript - and figures of around 40% for automated accesses aren't unusual.

[natto]

It could be possible that a lot of people are leaving a page before it has fully loaded. This will result in your Adsense ads not being loaded in time (hence no page impression). It does (but correct me if I'm wrong though) result in a page impression in server logs. The discrepancy seems to be too large and consistent for this but it's worth considering.

[adfree]

Would there be a theoretical potential for AS impressions only to count if the ad is really visible within the browser window and whenever your ad is off the window it does not count as an impression?

This might be stupid but theoretically possible, right? And it might be interesting for Google if compared to total page loads in comparison?!? If I was Google, I wanted to take this qualitative equation...

Jens

[adfree]

...and show it to the AdSense partner too in the stats...

[natto]

Hmmm, interesting thought Adfree. I don't think it's possible though but feel free to prove me wrong!

[figment88]

don't most stats packages filter out robots when reporting pageviews?

I'll go with the page not fully loaded theory. Depending on page structure this can be more or less relevant. If you do a 2 or 3 column layout with the bulk of info on the left and adsense on the right, you would be more likely to have less adsense impressions than pageviews.

This theory is easy enough to test, just put a simple page counter directly below the adsense.

[jomaxx]

yobo, you need to look at your raw site logs and count up
(1) ONLY the .html pages
(2) on which you are running the AdSense code
(3) that have loaded with status 200
and (4) show a referrer URL.
That number should correspond very roughly to AdSense's impressions.

If you don't have access to the raw logs, then I guess you could stick a snippet of Javascript on the same pages as the AdSense code, to write a transparent GIF image. Whenever that image is loaded, it should (again, very roughly) correspond to an AdSense impression.

[natto]

What are you counting as an ad impression? The display of one ad or the display of all ads served up by one instance of the Adsense code? (Which definition does Google use?)

For example, if you have the Adsense code in one page and it shows four ads, is this one ad impression or four?

[varya]

It's one impression regardless of how many ads are in the iframe.