How much of an impact is Twitter having in Iran?

This is a story about how I was caught gullible - twice.

Like everyone else who depends upon mass media for a fair amount of my news, I was a bit surprised that Twitter was arguably playing a significant role in the protests in Tehran over the recent presidential election. I mean, this is Twitter, whose model for communication opens itself up to breathtakingly easy parody, as Conan Seize-the-low-hanging-fruit O'Brien illustrates. (I love you Conebone.) Case closed... right?


Tonight I was prompted by a trading friend to look at an article from Pakalert saying that Twitter's presence in Iran is significantly overrated. It claimed that according to Alexa, it's penetration into Iran was almost nil. Interesting! I immediately posted the article on my Facebook page.

But something bothered me about it. After I posted it, I read the thing (talk about lack of due diligence!) and found the tone, ah, paranoid. Combine with the other linked stories, ranging from the generally accepted to tin hat-ish, I decided to dig in a bit more.

I first consulted the Alexa website (www.alexa.com), and took a bit of time to figure out what precisely the article was highlighting in its claim.



As it turns out, the article highlights that Iranians make up a very small percentage of Twitter users. Big surprise there. But Iran does appear in the top 23 countries. (It's 19th.) Alexa states that Iranian Twitterers make up 0.70% of the overall Twitter population.



Twitter also make it into the top 100 page ranking for Iranian websites (It's 97th).

Here's a representative sample of the top 100 sites in Iran by ranking, as of today, 06/20/2009:



If we want to see whether Twitter's impact WITHIN IRAN has been overstated, we need to wrap our head around the Twitter numbers for Iranians.

Unfortunately, those stats aren't easily accessible on Alexa, at least not from my crash-course self-guided lesson. But using the two data sets, I can get a sense of its impact.

In particular, we can get a handle on the percentage increase over May in Alexa, get the average numbers of unique visitors / page views from Compete.com, and combine that with the percentage of users of Twitter who use it from Iran to back out the number of Tweeters in Iran.

Methodology:

I picked up the relevant statistics from two web traffic tracking sites: alexa.com and compete.com.

Screenshots of some of the key statistics:





Note: I only have free account level access to Compete.com; consequently, I only have monthly data up to May 2009.

Assumptions:

There are a number of assumptions I make to estimate the percentage increase weighing appropriate for using the Compete.com figures from May to make an estimate for numbers in June.

Approach 1: Eyeball the average percent for May from Alexis.com, then use the most recent 7-day period to estimate present traffic.


Approach 2: To eliminate temporal effects (peaks seem to correspond to weekdays), estimte the fraction of users as of the 3rd peak in May, then calculate weight based on yesterday's numbers. (The idea being that yesterday/late last week was the third week on record in June, and will correspond roughly to the third week in May.)

BIG COMMON ASSUMPTIONS:

(1) Pageviews are proportionate to unique visitors for Iran - a conservative assumption if Iranians are using it significantly more than others to check in on what's happening, and perhaps a liberal one if they are either leaving it open and allowing Twitter to refresh, or aren't depending on twitter much at all.

(2) Unique visitor percentage is accurate for present circumstances. Unfortunately, no indication on the Alexa site on how they calculate the visitor percentage. (Trailing unweighted average of last 7 days?)

Results:






Conclusion:

It's tough to say how much of an impact Twitter might be having within Iran. A 97th page ranking isn't bad, but it's also not overwhelmingly impressive. Same for an estimated 180k-197k unique users per month, or 1.2 to 1.3 million pageviews. Assuming Facebook is censored, then it makes sense that students/protesters/others would not be using that. It'd be helpful to find a consolidated list of who is and is not voluntarily compliant with Iranian censorship restrictions. Twitter might in fact be the medium of choice, not only for its connection to SMS messaging, but because it's new enough that it flew below the radar.

In any case, trust but verify. Some have questioned the validity of Alexa numbers, I'm open to looking at other sources, should any readers know of a reliable alternative.