Hello. I created a report for our website and under the Browsers by Version section I'm looking at the Visits percentages. When looking at some of the browsers, I noticed the sum of the sub-rows percentages exceed the total for the browser. How can that be? For instance, Firefox shows as 12.83% of the total visits, but when you look at the breakdown per version, all the subrow percentages add up to 23.69%. The Othercategory appears to be what is throwing the percentages off, so maybe I'm not interpreting it correctly. Thanks.

Hello Janet,

I know it's been a while since you posted this question - sorry you didn't get a response. Just in case you're still looking for help:

I've looked into this a bit with some of my data and some of the data of the clients I support, and I've been unable to replicate the issue that you're describing. In all the cases I've checked, the number of visits for each version subrow, when added together, matches the number of visits for the corresponding browser. I've checked cases where the 'Other' category is present and cases where it is not, and in both cases, the numbers seem to add up consistently.

So I think it's safe to say that what you're seeing is a bit unusual - it's definitely not expected behavior from this report, and it likely indicates either something odd with your data or with your profile configuration. Are you using Webtrends On Demand, or a software installation? In On Demand, the issue would most likely be related to your profile settings (specifically, your sessionization). In software, it might well be the same issue, though we'd also have to rule out anything odd in the logs themselves.

The reason why I think your sessionization might be at issue is because the most obvious way to end up with the kind of numbers you're seeing is if Webtrends believes that a visitor's browser version changed in the course of a single visit. If I start my visit reporting my browser as Firefox 3.0, and a few hits later I report my browser as Firefox 3.5, both the "3.0" and the "3.5" subrows would count a visit. The main "Firefox" row would still count a single visit, so adding together the subrows would give you a number one visit higher than the main row. It's hard to imagine a significant number of visitors actually upgrading their browser during the course of their visit, though - we generally expect the browser version to remain the same throughout a visit. But if Webtrends gets confused about where your visitors' visits begin and end, it's possible that it could be rolling some visits together incorrectly, and in that case it's easy to understand how one "visit" might contain more than one browser and/or browser version.

Have a look at your profile settings under Analysis / Session Tracking, and ask yourself whether it's possible that the way Webtrends is defininig sessions might accidentally cause Webtrends to consider separate visits to be a single visit. For example, you might be using 'authenticated username' in an environment where different users might share a username, or you may have a custom definition set up that isn't specific enough for Webtrends to reliably distinguish your visitors from one another. If this is the case, changing to a more specific session tracking definition may help.

Of course there could be something entirely different happening as well. It's clearly not standard or desired behavior, though, so if you continue to see the issue and this information doesn't help narrow down the cause, please contact our Support group at 503-223-3023 and we'll help you identify a fix.

Regards,
Eric

Janet, If you're still looking at this item, could you supply a screen shot?

Hi Eric. Here's the screenshot you requested. I included it as an attachment but not sure I attached it correctly so please let me know if you can't view it. In reading through your previous response, I checked our session tracking settings and it looks like all of our report profiles use the 'Track user sessions using first party cookie' option. It is set as the default and also has the Param and Cookie options checked. I am still trying to understand WebTrends so am not sure how this affects the data. I plan to spend more time on understanding Webtrends analystics after the first of the year. I do know that our web server (apache) is what is generating the first party cookie. Thanks for your response and assistance with this.

When looking at the screenshot, look at the Firefox entries and notice that the total % of visits (if I'm reading it correctly) is listed as 9.95% while the Other total ALONE is 16.78%. Either I'm not interpreting it correctly or our data is wrong...??

12-29-2009_9-20-41_AM.gif (10 KB, 18 downloads) screenshot of browsers by version percentages

Apache is generating the cookie? Now I'm wondering if you should be using the "first party cookie" sessionization. That usually refers to the SDC first party cookie (yes, crummy labeling on WT's part).

What's the name of the cookie?

Well, other than the cookie issue ...

The screen shot really helps.

Sure looks to me like a WebTrends defect a.k.a. bug.

That last percentage looks like it has a decimal place error. For the data you have, it appears to be exactly 10x too big. Instead of 16.78%, it should be 1.678%.

I don't know which version of WT you're using but maybe you could check to see if the same thing happens in other profiles, or in other browsers for this same report. I'm still using 8.1 so can't check it myself.

Chris, I just now saw your emails notifying me of your response! (I have way too many emails) We are using WebTrends 8.5. and I did check another report for a different website and it is the same thing, and with more than one browser. I'm thinking the first party cookie is referring to the webtrends cookie, but I could be all wrong. I'll show my boss this thread of our correspondence on Monday and see what he says. Thanks.

I did notice that the Other line of the report was not included when I exported it as a PDF. ?? Not sure why.

Hello,

Recently on of our customer ask about this issue in this report. We opened a case with Webtrends Support, and they give us the answer to thas behaviour, it´s not a bug, but it´s true that the report leads to misunderstanding.

The explanation of this report, is that the primary dimension shows the percentage of the visit totals, and also the secondary rows "except" the "others" row. I mean, the version rows shows the percentage of the visits from the total of visits. But the "others" row show the percentage of visits of the rest of versions of firefox (in the screenshot). So We have

PRIMARY DIMENSION % visits from total
SECONDARY DIMENSION 1st row % visits from total
SECONDARY DIMENSION 2nd row % visits from total
SECONDARY DIMENSION 3rd row % visits from total
"Others" row % visits of FIREFOX from the rest of versions not showing above

So you shouldn´t sum the diferent rows of the versions plus the other row, because the percentages are not from the same source.

I hope that i explained correctly, but if it´s not clear, please ask again,

Regards

Hi Gfuente,

Thank you for your explanation. I'm still not quite clear on what the percentage from the 'Other' row is representing. Does 'Other' in this case represent other versions of Firefox not broken down on the lines above? And if so, what is the percentage a part of? 16.78% of what?

Thanks for your help.

Hello,

I´ll try to explain it with the data of the screenshot.

You have 52,322 visits of Firefox users, that are the 9,95% of the total of visits to the whole website.

Then the diferent versions of Firefox, also shows the % of the total of visits, in this case, the first version is 3.5.3 with 14,903 visits that are 2,83% from the total of visits. The same for the rest of versions.

Then, you have 8780 visits of Firefox, from "other" version not showing above. So you have 8,780 visits that are 16,78% from 52,322 of the total visits from Firefox. You can check it:
(8,780 x 100) / 52322 = 16,78 %

I hope that now it´s more clear.

Regards

Yes, the math does come out to 16.78%. Thanks for taking the time to explain it more clearly.

Gfuente,

Hi. In rereading your response posted 2/1, I just want to make sure I understand what you're saying...when viewing the percentages from the different versions of Firefox, those percentages are a percentage of the overall total visits from ALL browsers, correct, then the Other category's percentage is a percentage of JUST the Firefox total. Let me know if this is correct. Thanks.

Hello Janet,

That´s absolutely correct. That´s why it´s leading to confussion, because the percentages are from different things, so if you try to sum them all (to get the 100%) it doesn´t match.

Regards