Bookmarks

How many visitors bookmark your site

[Meike]

Forgive me if this has been hashed out before, I'm fairly new to Webmasterworld.

Has anyone ever done cross-site comparisons with regard to what percentage of users bookmarks their sites? I have noticed of late in the AWSTATS that the percentage of people bookmarking has been on a steady rise. I started paying attention back when it was 5% and in the last month or so it has crept up to 8.4%. My site is also only a few months old, so I don't have a long track record for data analysis.

So what I'm wondering is, what kinds of things can this bookmarking data tell us? I've come up with a few preliminary hypotheses with regard to the trend at my site:

1) Search engine results leading to my site are becoming more relevant over time to what users are searching for.

2) I am doing a better job at fleshing out my content -- enough to make more visitors want to come back.

Another note: Interestingly enough, much of my Yahoo search traffic disappeared with the latest update. When seen in light of my bookmarking data, this could mean that Yahoo searches were sending less relevant traffic my way.

I may be reading too much into the data and certainly am oversimplifying by leaving out other potential influencing factors.... Anyone else have experiences with this kind of bookmark analysis?

[bill]

Unless you have some sort of custom system in place the AWSTATS bookmarking figure is highly suspect. More likely than not you're seeing an increase in browsers and software requesting your favicon.ico file. That is not an accurate measure of the number of people bookmarking your site...and hasn't been for several years now.

[carguy84]

repeat what bill said, especially with the rise in popularity of Firefox.

[Meike]

Awww ... dang! I thought maybe I was on to something. Oh well. Thanks for the input.

[JAB Creations]

You could attempt to somehow detect favicon accesses from browsers that do not reqest favicons except when accessing via a bookmark. That may help you a little...but unless there is a helpful program out there you'll have to manually go through your access log! And that still may be a little off (but may give you some idea at least).

[cgrantski]

JAB has a good point. It's actually not too hard to capture bookmarking behavior (but not the behavior of coming to a site via a bookmark) by developing a filter the removes everything except IE 5.x and newer. Then, for this group only, determine how many visits (or better yet visitors) request favicon.ico. Take that percentage and generalize it to all your visitors, if you buy the assumption that IE visitors will bookmark at the same rate as users of other browsers.

You can get really useful stats by doing this ... not only trends, before/after redesign, etc but also which pages get bookmarked and how many visitors make more than one bookmark in a visit (supposedly highly correlated to site success). To do the "which pages get bookmarked" reporting you need a package that will tell you what page immediately preceded a favicon.ico request.

If you're into measuring quality of visits as well as quantity, this is worth the effort.

I should add that Firefox users find it much too easy to disguise themselves as IE users, and for some reason they think it's still a good idea to do it. So your data will be potentially contaminated by these users. We figured out how to detect these disguised users and don't count them as IE, but I don't know if it's worth going to all that trouble.

[j4mes]

You could attempt to somehow detect favicon accesses from browsers that do not reqest favicons except when accessing via a bookmark.

Correct me if I'm wrong, but isn't this exactly how Awstats works?

[cgrantski]

Um, yes, it is. The current version of AWSTATS uses IE browsers' requests for favicon.ico then generalizes to all visits using the ratio of IE to other browsers. Which brings us back around to Meike's original question, because AWSTATS seems to have nailed the method except for the Firefox-imitating-IE question.

I'd add to Meike's 2 hypotheses:

3. People in general are bookmarking more, web-wide(which is what he wanted to know when asking about what others had seen)

4. Meike is filtering out bots, spiders, monitors, and self-traffic better and better, and there is no actual increase in bookmarking behavior.

5. The number of Firefox users imitating IE has grown and there is no actual increase in bookmarking behavior.

Then there's the question of which version of AWSTATS Meike is using. Maybe old versions didn't separate IE from other browsers.

And I would have to wonder how much traffic we are taking about. With small samples, you can expect these numbers to wander all over the place just by chance. A month is a really short time to see a trend like that unless Meike's been implementing some improvements that plausibly could make those numbers climb.

[larryhatch]

I've wondered the same thing for a long time.
Unless there is some special indication in access_log records,
I don't see how any of these packages could give reliable numbers.
All I see is "bookmarked or typed" which tells me nothing really. -Larry

[bill]

We've had this thread before. ;)

I hope you're prepared to filter out all the other software that now uses the favicon.ico file like RSS feed readers and the like. FireFox isn't the only software to use this.

There are certainly ways to track bookmarking behavior, but I'm not convinced that all this reliance on the favicon.ico file is worth it. If the screenshots I see for IE7 are correct, the next version of IE uses the favicon.ico file on its tabs the same as other modern browsers.

[cgrantski]

Heh, I think we've had most of the threads here before! But they get better each time of course ;-)

Bill those are important points, and can you possibly give a quick summary of how to recognize RSS feeds is server logs? This is something I've never checked into. I assume that if a site doesn't offer an RSS feed, there will be nothing in the logs ... or will there?

[Meike]

Thanks to everyone for all the input. Now I can stop paying attention to that stat and wondering what it means.

It sounds like -- at least for now -- figuring out bookmarking behaviors is more trouble than it's worth, particularly while I still have a day job I have to spend time going to. It sounds like my Web time might be better spent producing more content that is bookmark-worthy!

[bill]

can you possibly give a quick summary of how to recognize RSS feeds is server logs?

This gets difficult because you'd have to be on top of all the major software and their user agents. Readers like FeedDemon are nice because they tell you who they are in the UA, but not all of them are as forthcoming. Compiling and maintaining a list of this software might be prohibitive given the ROI.

Then you would have to keep track of all the web based services like Bloglines, that aggregate feeds and show the favicon.ico from your feeds regardless of the browser. The online services vary the times that they come and check your files, but that will also play into the number of times favicon.ico gets pulled.

cgrantski already knows this, but I tend to side with Meike's thinking, "It sounds like my Web time might be better spent producing more content that is bookmark-worthy!" ;)

[cgrantski]

That's too hard! I like the easy life of the web traffic analyst! :-)