Since announcing Twitter Karma on December 29th, it became popular pretty quickly in the Twitter community. I watched the traffic come in from various sources and being interested in this sort of thing, I wanted to know exactly what kind of impact it had on my overall traffic.

Using Google Analytics and Site Meter, I’m able to keep an eye on my traffic and get some useful metrics about it. Lets start with a graph of the past 7 days worth of visits (yellow) and page views (orange) from Site Meter:

What’s immediately interesting to me about this graph is that, previously, my weekend traffic was usually lower than my weekday traffic. However, the Twitter Karma traffic peaked on the weekend, which surprised me.

To get an idea of the number of users it brought me, I looked to Google Analytics. Here’s the Absolute Unique Visitors data for 2007-12-28 through 2008-01-08:

5,540 Absolute Unique Visitors

Friday, December 28, 2007	2.73% (164)
Saturday, December 29, 2007	3.36% (202)
Sunday, December 30, 2007	2.36% (142)
Monday, December 31, 2007	5.72% (344)
Tuesday, January 1, 2008	4.66% (280)
Wednesday, January 2, 2008	6.92% (416)
Thursday, January 3, 2008	11.84% (712)
Friday, January 4, 2008	9.68% (582)
Saturday, January 5, 2008	21.75% (1,308)
Sunday, January 6, 2008	16.58% (997)
Monday, January 7, 2008	9.06% (545)
Tuesday, January 8, 2008	5.35% (322)

To separate out my regular blog traffic, I pulled the Top Content report for the same timeframe:

291 URLs were viewed a total of 10,159 times

Page Title	Pageviews	Unique Pageviews	Time on Page
Your Twitter Karma	5,653	4,494	00:03:13

So, of those 5,540 visitors, 4,494 of them went to the Twitter Karma page, or close to 81%.

To be honest, ~4,500 visitors isn’t a big number at all. What makes me really curious is finding out how many “active” users the Twitter service has. What percentage of that total does 4,500 users represent over that time period? Less than 1%? 5%? 10%? 50%?

I’m sure the Twitter guys aren’t going to open their kimonos that wide and share that kind of information with us, but I wonder if there’s any way to collect that information anyway. I guess I could sample the public timeline every few seconds for a few weeks, then compute the uniques and update frequency to come up with an estimate. This won’t take into account those who have their updates protected, but I’m willing to bet fewer than 20% of Twitter’s active users have done that.

Still, this was a fun application to hack together and it seems to be well-received amongst the Twitter community, so I’m really happy with it. I hope someone can come up with another great tiny app. idea for Twitter that I can hack on, next.

Tags: Twitter Karma, Google Analytics, Site Meter

The weekly Hackfest took a bit of a break due to the winter holiday and resumed this past Monday. The goal for this session was to turn the visual design created last time into actual FBML that would render inside a profile box.

The first step was to create the database schema, which was mostly outlined in the design document. I went ahead and created actual DDL scripts for MySQL based on what was described in the design doc. I’m expecting these to change as development progresses, but this represents the minimum required to implement our current visual design.

The general approach for the profile box implementation is to use profile.setFBML to set the FBML of the profile box to simply “<fb:ref handle="profile_box"/>“. We want to do this because every day when the profile box needs to be updated for every single Facebook user who has added our application, we don’t want to have to invoke profile.setFBML with the newest information–that just won’t scale well. Having the profile box simply contain an <fb:ref> means we can effectively update everyone’s profile box with a single call to fbml.setRefHandle, once per day.

I’ve also created a SQL script that can be used to populate the MySQL database with sample data for testing. The only MySQL-specific part of it is the use of “CREATE TABLE ... LIKE ...” which allows us to use the submission_template and votes_template tables to create our individual instances of those tables for each list_id, rather than embedding the DDL for each of those in our code itself.

After all this, here’s a screenshot of the profile box off my own Facebook profile page, pulling the sample data out of the database and rendering it via FBML:

The rating widgets–the “thumbs up” and “x” icons–don’t actually work yet, but that’s going to be next week’s challenge, getting the Facebook “mock AJAX” working.

Tags: Hackfest, New Jersey, Facebook, MySQL

@Debra_Hamel asks,

“Is there some way to ask TwitterKarma to look at a second account once you’ve input the details for one account?”

The simplest answer is to close your browser completely, and the next time Twitter Karma tries to access Twitter, it will prompt you to re-authenticate. Of course, that’s no fun–I’d hate to close my browser and lose all my tabs, so I looked for another way.

In the Twitter API, there’s a “verify_credentials” method that, coincidentally, seems to change the effective user of the browser session if authentication is successful. You would invoke it by navigating to an URL like this:

http://USERNAME:PASSWORD@twitter.com/account/verify_credentials

Replace the USERNAME and PASSWORD with your Twitter.com account information. You should get a plain page back that just says “Authorized.” After that, reload the Twitter Karma page and click the “Whack!” button again. It should now load your friends and followers for the other account.

Tags: HOWTO, Twitter Karma

So, I finally let the cat out of the bag a few minutes ago and announced Twitter Karma publically.

So, what is it? The other day, @StephAgresta said, “Twitter desperately needs page navigation on followers / following list. Also sort functionality by type (reciprocated or not) is a must.” I suggested a simple mashup that implements this and started hacking on it. Three days later, it has enough functionality that Stephanie said she thinks I should release it publically.

Basically, it’s a Flash application that fetches your friends and followers from Twitter when you click the “Whack!” button, then displays them for you, letting you quickly paginate through them. By default, the list contains all your friends and followers and is sorted by last update, showing those who most recently updated first. You can sort the list alphabetically either ascending or descending by Twitter ID. You can filter the list in several ways: only friends or only followers, all friends or all followers, and mutual friends.

It’s not meant to be a full Twitter client–there’s plenty of those already and that’s not a particularly interesting or challenging problem to solve, anyway. I’ll be adding a few more features to Twitter Karma soon, though. If you think you have a must-have feature idea, go ahead and let me know about it in the comments.

Tags: Twitter Karma, Flash