Russell Beattie

The great thing about having your own non-commercial website is you can mess with it without much concern if you do something wrong. Since I've got a decent amount of traffic and search engine placement, I can tweak it to get a better idea about how people use the web. I've been messing with the design of the site lately, playing with both CSS and JavaScript (check out my spanky Ajaxy Archives page for example) and I've also been looking at the click paths of users to see what they actually do here. I've done this before with RSS - adding forms, comment links, etc. and now I'm playing with the main site a bit. Check out my StatCounter page so you can play along at home...

Some of the things I did recently: First, I shrunk my header down to just a bar across the top page and widened the body to 765 pixels. This has the effect of putting more content on the screen "above the fold" and pulling the next and previous links from the bottom of the page up as well. I messed around with the fonts and colors, and came back to my favorite black on white and Arial font, which sort of conveys simple and clean (or 100% generic, depending on your viewpoint). In a world wide web of hyper designed pages, a clean white page seems to stick out. I also added a Search bar at the top of the screen, and changed the previous/next links to be short labels rather than the long titles of the posts as they were before.

Okay, so now step back and think like someone arriving to my site. In order of popularity, they come from 1) Search engines 2) Links from other sites 3) RSS Aggregators and last 4) Bookmarks. The first two are actually related because many times I can see that searchers have "hopped" here from another blog after they found that site in their search results and continued looking. RSS links are harder to detect, but are generally obvious because of their lack of referrer or view of a specific permalink which is still in my feed, and then don't continue to explore the site. The last are the old school folk who usually bookmark my front page. Again, I can see this by noting that there's no referrer and they come right to the front page, not somewhere inside.

So what I'm really focused on right now is the searchers and kind of noting what they do. I've seen a huge increase in the continuous page views since I changed over to Previous/Next labels, which makes perfect sense. If a searcher has too many links and too many options, they kind of click around for a sec, get lost and go back to what they were doing. Given a clear path, they seem to tend to follow it if nothing else catches their eye.

Looking at the viewer paths, it seems to be like this: A searcher arrives here from an engine (usually Google, where I have the most juice), and scans the document. They get to the bottom of the page, see the two bold links, and choose one. They either are determined and continue in one direction or another indefinitely, or they're sort of wishy washy, and seem to want to view *both ways at once* by clicking back to the original page and then again in the other direction, and then maybe going back again. Some searchers will then click on my front page to see what the latest stuff I've said is, and then on to my About page to get more info about the jerk who's writing all this bullshit, or they'll just sort of fade off from a page, probably closing that window or tab and returning to their original search. Here's a recent example:

Bangladesh, 0 returning visits

So looking at the table from top to bottom, you can see that this is someone who's never been to my site before, came from a search, spent a few minutes reading S60 app reviews, then clicked on the Next link a few times, and was gone a minute or so later. At least it seems that this person spent a 30 seconds or so on the first couple of pages after his original search, I can see others come in and rip through 10 or so pages in just a few seconds.

This brings up my first thought: how useful are the next and previous links really? Well for people who arrive from aggregators or bookmarks, they're actually quite useful. I won't copy it here, but there's another visitor path just recently where someone arrived at a recent post, and then seemed to "catch up" by going back a few links, and then forward to the latest article. For them the next/previous links are really useful. For the searchers, however, they're pretty useless. The person above came looking for free Nokia software, but within a few clicks was reading some personal stuff which had nothing to do with mobile phones. You can see this quite a bit - searchers arrive, click around for a bit and leave.

This makes a feature which is really big and bold at the bottom of the screen useful for the least number of people coming to my site. Hmm. You could make the argument that those are my regular readers, and thus are more important and I can see that. But obviously, for the casual visitor, next and previous don't do much for them.

This is where tags or categories would've really come in useful, but since I've never gotten around to adding them, I've done a couple things instead. First - I integrated Yahoo's Content Analysis API at the bottom of each individual post page (I use WP-Cache, so that should help on the daily query limits). I pass a copy of the post to Yahoo! and get back a bunch of relevant topics which I list under "Similar Topics In This Site". THis is good, these aren't "tags" per se, so I'm not really presenting them as such. Instead, they link to my Search page instead, which will bring up a list of posts about that specific topic. To make sure the searches were good, I used the Yahoo! Search API instead of WordPress' default for more accurate and contextual results.

The end result is actually pretty cool. Someone arrives at a post on my site, reads/skims it, gets to the bottom and sees a list of items similar in topic to what they were reading, they can then click one and then see a bunch of content I've written about that topic driving them deeper into the site. Instead of someone having to "guess" at interesting things to read about on the site and then searching, they are now given a set of clean choices instead. For the searchers, this means that instead of arriving at a page, reading it and then clicking around futiely in one direction or another getting random content clustered around a certain date, they can choose from a list of items instead which are similar, and more accurately reflects their interests. And no this isn't a promo for Y! Search stuff... I was just trying to figure out a problem and realized I had some interesting tools at hand. I even used the PHP serialization stuff... wow, that makes life a lot easier than parsing XML. :-)

I've tried it for a few hours now and it seems to be very popular (maybe too popular - I'm going to easily burn through my daily limit...) where searchers used to click on the next or previous links because it was the only thing to do, they're now shunted to the search page instead, and I've seen people either click through to another topic that interests them, or try a few other searches as well. That's great.

We'll see how it works out over time...

-Russ