How Can a Page Have More Unique Pageviews than Total Pageviews?

[glennfriesen]

The question and the attached image say it all... how is this possible?

2v1uo8w.jpg

[glennfriesen]

tool: the asynchronous tracking code from Google Analytics

[glennfriesen]

It seems very improbable, but it's possible (not sure how I'd check for incognito visitors... I'll try to figure this out sometime in the next few weeks).

I tell you what - this specific issue itself is confounding, but the bigger problem is the small but evolving perception that all of the analytics data behind my advice is untrustworthy (a client's growing suspicion that "if this GA data is wrong, then all GA data must be wrong; and Glenn's advice, being based on this data, is invalid". I can just show all the new rankings we get for proof in the pudding, but I just personally want to figure this out theoretically to eliminate any remaining buried seeds of insecurity.... lol!

[KeriMorgret]

I've passed this one around internally for ideas too. Is this in a subject area where many people might be visiting in incognito mode? It is a bit of a head scratcher!

[glennfriesen]

The client never noticed, but is about to on an update report (see attached).

Still happening, still baffling... really hoping someone can help me figure this out... several dozen people I've discussed this with don't believe me, see this, then are baffled too --

voNwe.png

[KeriMorgret]

I haven't seen this before myself, and just now saw this question. What happens if you set the date period to less than a month? Do the numbers look better then? What about if you show things by day instead of week -- is there a day that is out of whack?

Did the client ever notice this and ask?

[glennfriesen]

Thanks - that makes a lot of sense, holycap.

Yup - This report is not for a custom segment, and is beyond a month.

Right again, this image is for an infrequently visited page -- so that low volume could lead to bigger margins of error. I would expect those margins of error to match, though, for pageviews and unique pageviews - wouldn't you? Why would such similar metrics have such varying margins of error?

[glennfriesen]

Mad props for replying --

This screenshot is from Google Analytics. My server logs, and other analytics software (clicky) don't report the same discrepancy. Since the owners of this website only check GA, I just wanted to figure this out so I can answer their question if they ever ask).

The code is installed correctly. It hurt a little to even consider that theory - though I did review the site firsthand and check....thanks for keeping me humble, tho! This site's bounce rate is about 12%, so I'm pretty sure it's not because of the visitor pressing the backbutton before the page loads (pageload lasts 2 seconds on average, and 95%+ of visitors spend over 10 seconds).

Still unresolved.... and this question is one of the few that seems to stick with me, since this one bizarre example makes all of GA seem less trustworthy. Lemme know if any other ideas spring up ---- Until then, I'll be like the attached images. 
Thanks again --

hands_twiddling_thumbs_lg_clr.gif image11.gif f923589989ee7d2c47f9106d3387db29b9b2278.gif

[dealzon]

Another thing he could consider is the fact that this is an Analytic pull of a date range that looks to be beyond a month, so if this is a custom segment, it might be generating a fast access report that's known to be inaccurate in its ballpark estimate at times, especially when numbers are this low, the margin of error can be much bigger.

[JoeAmadon]

Magic. Obviously.

I can't give you a definite answer but I imagine what is happening is that the data points are being pulled two different ways. Perhaps unique pageviews are being pulled from requests for a URL but total pageviews are being pulled from tracking code and the code isn't loading because the code is put in wrong or the user hits the browser's back button before the page loads.

Which tool are you using?