Liwen Vaughan, University of Western Ontario

[Please note: These are merely my notes on the presentation, taken live while the presentation was in progress and edited for sense afterwards. They are not a verbatim transcription of the presentation, and any errors are mine. Please contact the researcher directly for more information on her work. — JRD]

Partial notes on the presentation:

There appears to be a correlation between high Alexa rankings and big/high performance/profitable companies. (Not claiming causal relationship, but the purpose is not to establish causation but to use correlation for prediction … business performance measures aren’t always available, so “traffic to company Websites could be an indicator of the company’s business performance”).

Given the premise, how could Alexa data be “crunched” to create competition maps of different companies?

“Large scale web traffic data” could be useful for predicting business performance.

Limitations: validity and reliability of Alexa data unknown … one industry only (in this study). In a previous study they started with one industry and then extended to other industries.

Future study:

Monitor web traffic data over time to examine reliability; find other web traffic data to triangulate with Alexa data; test the method in other industries. One limitation is that other traffic data can be hard to come by.

Might pay for some data on a particular company’s web traffic in order to look at Alexa’s reliability. Once Alexa is “validated” they would like to analyze content data and web hyperlink data to give a more complete picture … overcomes the limitation inherent in using only one data source.

Question from the audience: Did you consider looking at one week’s data at different periods of the year in order to look at seasonal variations?

Answer: You can see that in a company like Walmart, where traffic peaks around Christmastime, and slows down in July. We would like to look more at seasonal variations.

Question from the moderator: Are you looking at query content analysis?

Answer: We aren’t looking at queries at all. Queries look at people’s interest in the content, we are looking at traffic rather than people’s interest in the content.