Don’t Believe – Test It

Recently I got a blogging advice about how to optimize your blog:

Keep your blog size as small as possible: 15 kilobytes is fine, 10 kilobytes better. Remember: most of your visitors will use 56k modems to read your blog.

I thought this hint to be a good piece of an advice and started to implement it: I took away some useless CSS code from my blog and limited the number of displayed posts to 5.

After this I asked myself: “Is this tip really useful?” I went to one of the popular blogs I knew and checked the filesize: the main page was almost 80 kilobytes (and the feeder was about 85kB). I decided to take my previous 10 posts per page back. If the top popular blog is big & popular, then the file size won’t be a limitation to me neither (I also thought about my audience – game developers – while deciding) .

It’s always good to listen to someone’s advice, but you should test the recommendation before accepting it.

Don’t take my word for this – test it.

2 thoughts on “Don’t Believe – Test It

  1. @Ken: That’s true – the example I give here is not the best one. But I still say that the site I checked is getting more than million visitors yearly and is not the first “long” blog I’ve seen. It was kind of a test of “one site suggesting small against one popular site suggesting big” test with my perception (true – you shouldn’t use “opinions”) about the general size of blogs (gamematters, gbgamer’s etc.) – and as said: I remember my target audience (game developers).

    Very good comment. To make a proper test needs a systematic approach on how to conduct the research/test situation and how to evaluate the results.

  2. Checking one site isn’t really much of a test. A popular, established site will be able to bend or break many rules. Excessive size could keep a smaller blog from ever reaching higher. So it’s good to test, but you need to do a real test if the results are to be meaningful.

    But you also have to keep in mind your target audience. Game Developers will tend to have better technology in general, so I’d bet that fewer of them would have 56k than compared to the general public. But then you should really test that theory. :)