Several people have said to me that technoranki should take into account blog traffic as one of the factors used to calculate blog rank. I’m inclined to agree, but have held of this far for two reasons:
- It would require people to have a tracking image on their site
- It could put too much load on this server, bringing all the sites down!
I think I may have a fix for point two, so the question is: would people put the tracking images on their sites? To make this kind of measurement a factor in the blog ranks I need a majority to be in favour of it. Using traffic as contributory factor to rank will clearly bias that results in favour of those with the tracking code, so before I do this I’m keen to hear what other people think.
Using tracking images will allow you to display real-time icons of your ‘blog rank’ (like the Google PageRank) on your blog, and this may add an additional interest factor to the service. The icon will be the same size as the current BritBlog icons, so should be easy to integrate with your sites.
So what do people think? Personally I’d like to try this. If the traffic is too much we can always look at getting another server online. If we go ahead with it, I’ll need a bit of time to set it up, and then it will probably take a week or so for the statistics to sort themselves out.
January 23, 2006 at 3:52 pm
Been thinking about this, and it looks like it shouldn’t be as big an issue as I thought it might. Processing huge amounts of data each night takes up system resources, so I was worried this may cause some issues too. However, it looks like I can process the results incrementally throughout the day, so this should minimise the impact of the rest of the web site.
January 23, 2006 at 6:03 pm
Personally I would have no issue with putting a tracker on my site.
Could you just confirm though that traffic would be one of a number of methods that blogs would be ranked on.
For the short time I have been here it has been interesting to note fairly regular changes on the top blogs list – it would be good to keep them a shifting list – rather than risk the same names dominating for weeks on end if just traffic was used as on some other sites?
January 23, 2006 at 6:26 pm
Jem, you are correct. Traffic measurements would only contribute towards the final blog score/rank. The ranks are calculated nightly, but the traffic component used would be based on the last 7 day’s worth of traffic. This will help smooth out daily bounces, but will still allow sites that have spikes in traffic for legitimate reasons to be noticed.
January 23, 2006 at 7:17 pm
A rolling real-time chart, based on the last 24 hours traffic, say, would be cool. It’s important to get some movement, or the chart will become boring.
I think there’s a danger this will be very complicated and, worse, meaningless. Today Plasticbag is number one. Good for them. But number one at what?
January 23, 2006 at 7:38 pm
I too would have no problem adding tracker code to my site, if need be. I already have a BritBlog icon on my site, so a tracker of a similar size wouldn’t be too intrusive.
January 24, 2006 at 12:12 am
Yes, I would have no problem with a small chicklet either.
January 24, 2006 at 11:20 am
I too would be willing to put a traffic image on my site … obviously it will depend on how well the server can cope with the additional load, and if you believe it can then great!
I still think that other/existing methods should also be employed though to give a more balanced listing.
January 24, 2006 at 11:23 am
Excellent – well it sounds like this is a road we should go down.
Andy: the ranks won’t be based soley on traffic – there are several other things that technoranki measures in the background…
January 24, 2006 at 12:44 pm
Traffic is an “interesting” measuring tool because you can measure it in several ways. For example, adding a tracker image is ok, but many people have security settings switched so that images are only loaded from the same server that sent out the web page. This means certain types of audience won’t get registered, so for example, a brit blog that deals with computer securoty is unlikely to score well because the majority of it’s readership will be blocking non-local images.
It’s also necessary to consider image cacheing, and specifically, proxy image cacheing. It’s common for ISP’s to cache images at a nearby point-of-presence, so you need to specifically ensure that the image cannot be cached, perhaps by giving it a dynamic name or setting the HTTP header appropriately.
Assuming you can get around these obstacles; there’s still the issue of how site design affects the number of images that are loaded. If a BlogA limits discussion to 10 per page discussion and BlogB opts for blocks of 100, then assuming their readership is the same, BlogA will be hit 10 times more often than BlogB. You can’t get around this by demanding that the image only goes on the home page because the majority of hits that come from search engines never end up on th ehoe page, so traffic would be lost.
In a nutshell, analysing traffic can be useful to confirm that a diverse selection of the worlds population is visiting a site, but there are many more issues than I’ve mentioned here, which prevent it from being a good measure of popularity.
What a lovely paradox: High Traffic != High Popularity.
January 24, 2006 at 1:41 pm
Personally I think that blog traffic has to be *the most significant* indicator of a blogs success/popularity.
At the end of the day, if people don’t read what you’ve written… what’s the point.
I don’t know what other factors you are taking into account – but if it’s technical features like tags etc, then these say nothing about the quality of the blog. In the end people vote with their mice.
January 24, 2006 at 2:03 pm
I concur. Tracking device would be no problem atall.
January 24, 2006 at 2:32 pm
Rich: You’re quite right – this was the driving force behind technoranki – moving away from pure traffic analysis to something a little more thought out.
All the problems you outline are issues that dogged the old blog charts, which is the main reason I didn’t restore the charts when we moved servers.
I think I have a reasonable solution that involves just looking at unique visitors within a given time span rather than looking at how many pages the visitor looks at while at a blog.
Caching issues can be taken care using the correct headers as you suggest too, but this still will not prevent the overzealous caches from getting in the way. At least this problem will be seen everywhere – hopefully in equal proportions.
The only remaining issue is that of the (overly) security conscious as you cited, but this can’t be helped!
January 24, 2006 at 7:59 pm
WIll it or does it take into account feed reads and subscriptions too?
(This area has always puzzled me because i use sitemeter which doesn’t take these into account making my readership look smaller than it probably is.)
I have no problem – with adding the icon – I already use “BlogTopSites” which I think does a similar kind of thing.
January 24, 2006 at 8:40 pm
Sam, it won’t take feed reads in I’m afraid – at least not yet!
Technoranki works in quite a different manner to BlogTopSites. BlogTopSites measures “unique visits each blog receives every week”. Technoranki will make use of this type of information, but looks at several other aspects too (which at the moment need to be kept a secret I’m afraid!)
January 26, 2006 at 3:55 pm
Me, me! I’ll test it. Pick me!
How about we only have a testing base of 50 users, and I volunteer to be one of them. That way I have a guaranteed place in the top 50 – I’m not going to make it otherwise.
Even if we don’t use this method of testing I’d still be happy to be a guinea pig.
January 26, 2006 at 9:48 pm
oh – knew about the technoranki’s “severa; oother aspects” I was just equating the tracker to topblogsites – ie the icon you would like us to install would read the number of visits and then you would take number and add all sorts of other factors to give the result?
January 27, 2006 at 3:38 pm
All this talk centres around measuring hits. I noticed that some stats I already use on other places tell me “How long” someone was on a page.
Think about
1. a site with a good layout when Technical information is presented.. two page changes (2 hits?) & 5 minutes reading…
2. A poor site with Bad layout. I flick through 10 pages (10 hits?) in 20 seconds & decide there’s nothing worth reading
Point is, especially with Blogs! It’s how long people spend reading that is the measure..
Is my logic & understanding of the process wrong here? Further, Is it possible to determine the address of the hit so that a visitor only gets 1 hits per day, no matter how many times, or hits they registered?
January 27, 2006 at 7:22 pm
We had a tracking code before for top of the blogs, so it shouldn’t be a problem.
January 27, 2006 at 7:52 pm
Barry, thanks for your thoughts. I know what you’re saying, but I don’t think it’s practical for us to measure that. I don’t know how technical you are, but this kind of measuring would require the use of cookies originating from technoranki. A large chunk of web browsers reject third-party cookies (i.e. cookies coming from a domain that isn’t the same as the domain of the site that the user is visiting), so this would introduce a lot of misleading data.
Another approach is to log a blog reader’s IP address, but how do you know that it is one user reading the site and not an office of two thousand staff behind a firewall?
The point is that an accurate measurement of traffic is unnecessary. All we need is to measure traffic from sites in a fair and consistent manner. I should also point out that technoranki won’t be counting page impressions anyway, but looking at ‘other stuff’ relating to traffic.
Sorry for keeping some mystery about how technoranki works, but we need to keep the details fairly private in order to limit the possibility of people abusing the system…