Further on bot

Thanks to the many people who e-mailed or commented regarding my bot request. Rather than reply individually (yet), I figured I would explain more about what I was thinking of.

Tumeke, Open Parachute and Half Done all publish ratings of NZ blogs. A great service they provide, and I hate to think how much time it takes up.

The Tumeke ratings are based on:

1. Daily visits
2. Technorati links
3. Average posts/week
4. Average most comments on a post/week

Half Done ratings are based on:

1. Alexa global rating
2. Alexa NZ rating
3. Technorati authority score (no of incoming links)

Open Parachute ratings are based on:

1. Visits/Month
2. Page Views/Month

What I have in mind is a ratings site which combines the best of all three sites using a bot that could scan all NZ blog sites and pull up data for each blog from publicly available sources so that people don’t have to spend hours and hours producing the ratings. What are the criteria I would use:

Visits

1. Total Visits for Month – this is the hard one. This data can be reverse engineered from Alexa but Alexa isn’t the most reliable. If sites use Sitemeter, then this could be automatically read, and Alexa is no Sitemeter.
2. Total Page Views for Month – as above
3. Total Unique Visitors for Month would also be useful, but unlikely to be available without access to Google Analytics for a site

Links

1. Alexa Sites Linking In
2. Technorati authority score

Posts

1. No of posts in the month

Comments

1. No of comments in the month
2. Average comments/post
3. No of unique commenters

One could then develop a formula along the lines of calculating each stat as a percentage of the average for that stat over all blogs. So if the average no of posts in a month is 50 and you post 150, then your score is 3.0 (or 300%).

Finally you’d weight the criteria. I’d suggest:

1. Visits 40%
2. Links 25%
3. Post 20%
4. Comments 15%

One could also weight the sub-categories if you really wanted to.

So the idea is that one could just feed a list of sites into the bot, and hey presto it would dig up the raw data which could be exported to Excel, where they could be converted to scores and ratings.

I don’t have a budget for this, as it is non profit. So my preference is bored geeks wanting a challenge. If none of them exist :-), then I can probably throw some dollars at it, but it won’t be a lot.

Feedback is also welcome on the proposed data to be collected, and suggested scoring and weightings.