|Analytics vs. local count, WAY off|
On April 25, I changed from Adsense to a more specialized ad network. I've noticed an unusual drop in traffic since then, and when I compared the year before 4/25 to the year after 4/25, it looks like the shift in traffic began at exactly the same time as the change in ad networks! Literally, the same day.
I found that to be a pretty big coincidence. The load time is the same (within 0.3s, anyway), and no one has said anything about having a problem (which they normally would), so I'm not sure that this can be blamed on the ad network. But I'm talking about a 20% drop in traffic, which comes to about 5 million pageviews since 4/25.
But it turns out that my local count is almost double of what Analytics says! For example, on 8/14/13 my local count shows 223,707 pageviews, but Analytics shows 111,481. On 8/13, my local count shows 202,344, but Analytics shows 122,480. 200k would be closer to what I would expect, based on last year's reports (last year, I had around 190k pageviews on these days).
I could easily accept a 10% tolerance for robots, etc, but 50% is a little harder to swallow.
I've checked during off-peak times to make sure that a pageview only counts once in my MySQL script, and it seems right. Of course, if it were mistakenly counting each pageview twice, the count wouldn't end in an odd number, anyway. I wasn't tracking locally before April, so I don't know if it was closer to accurate then.
I can't help but be suspicious of the big coincidence that Analytics shows a big drop in traffic immediately after changing ad networks, but my local count doesn't. If this were any other company, I would totally suspect them of showing false results in order to make me think that my site was suffering due to the change. But with it being Google, I wonder if the coding for Adsense and Analytics may have been intertwined in some way?
So my questions here are:
1. Has anyone noticed their traffic in Analytics dropping unexpectedly after leaving Adsense?
2. Has anyone compared a local pageview count to Analytics, and found Analytics to be off by this much?