> Q: What is Akamai IO?
A: Akamai IO is a "destination site" that demonstrates Akamai's unique perspective on the Internet, as well as acting as a centralized home for Akamai's technical community efforts. The primary objective of the site is to make the Akamai Intelligent Platform "tangible", so to speak, through data and visualizations derived from usage of the platform by thousands of enterprise customers and millions of end users around the world. The "Akamai IO" name is derived from "Akamai Internet Observatory", which was the project's working name.
> Q: What is the source of the data?
A: The data feeding Akamai IO comes from sampling traffic of most of the websites delivered by the Akamai content delivery network (CDN). This sample includes over a billion requests each day, sampled from across the world.
> Q: Does this data have some sort of notable bias?
A: The only bias in the data is that it’s measuring traffic from Akamaized websites, which tend to be websites with a relatively high traffic volume. However, since the data we’re showing relates to the users browsing the websites, not the websites themselves, this is probably not a concern. Note that the data collected until mid April was from websites catering primarily to a US audience, and thus did have a bias.
> Q: Why do the browser usage percentages shown in Akamai IO differ from those published by [source]?
A: Browser market share data depends on both the source of the data and the way it is interpreted. If Akamai IO shows different results than those shown elsewhere, compare the source of data and method of client identification used to decide which is more applicable to you. We also recommend you combine this information with data from your own web analytics where possible.
> Q: Why does a browser appear in the pie chart view, but not show up when I switch to the line chart view?
A: In the pie chart view, the percentages displayed are averages over the whole selected time period -- if the average is large enough to place the browser within the top 5/10/15 (whatever quantity was selected), then it is displayed as a member of the pie chart. However, in the line chart view, the determination for inclusion is based on the browser's percentage on the last day of the selected time period. If that percentage is still large enough to place the browser within the top 5/10/15 (whatever quantity was selected), then it is displayed as a on the line chart. However, it falls outside of the selected quantity, then it is subsumed into the "other" category.
> Q: Does Akamai IO count page views or requests?
A: Akamai IO statistics are based on the number of requests made from each user agent. In future, we plan to show these statistics by content type, allowing a more refined drill down into the data.
> Q: How do you segment "cellular" vs. "non-cellular" networks?
A: Akamai is segmenting networks into granular subnets, and analyzing the traffic seen from those subnets. Specific thresholds have been established for the number of requests, number of mobile user agents, and overall percentage of mobile user agents - if traffic on the subnet exceeds the thresholds, then it is classified as "cellular". We periodically evaluate the networks and the detection methodologies, and adjust the thresholds to tune for more accurate detection. In addition, we have also seeded this classification process with other network-related data - autonomous system names will often indicate that a network is being used for cellular (mobile) traffic. Note that the classification process is best effort, and due to the fluid nature of these network, may not be 100% accurate.
> Q: How do you identify the specific devices?
A: Our mapping of user-agent to a client ID, as well as most of the metadata about the devices, is based on WURFL, with various customizations and edits made by us. WURFL is a great repository of devices, and helps us both in identifying the client device and in understanding its properties. WURFL is provided by Scientia Mobile, who is providing amazing value to the web community by supporting projects like Akamai IO.
> Q: What is the roadmap for Akamai IO?
A: Currently Akamai IO includes statistics on browser usage, and related trends, on fixed & mobile networks. Going forward, we plan to:
> Q: What happened on Feb 16, 2013? Why do the data look so different compared to the past?
A: On Feb 16, 2013 we've switched to a new global data stream, which provides more sampling data with improved accuracy. We’re still working out some of the kinks in this new data stream, trying to understand various weird patterns, but overall the new sample set is broader and a better representation of the internet.
> Q: What differentiates your data from that available elsewhere?
A: Akamai's customers include some of the largest sites on the web. In addition, we see nearly two trillion requests for content a day from users in hundreds of countries around the world. While extracting information from a data set that large is a non-trivial challenge, sampling even a fraction of a percent of those requests provides us with a data set that remains statistically valid. In addition, as it represents active real-world usage, it provides unparalleled insight into actual usage trends and user experiences.
> Q: Do you have data on [topic here] and/or plans to incorporate that data into Akamai IO?
A: We just might. Please send us a note with additional information on what you are interested in.
> Q: Why isn't Akamai involved in [community project here]? I have a community project that might be relevant to Akamai - who can I talk to about it?
A: Please send us a note with additional information on the community project, including a link to the relevant website, as well as comments on why you believe that it may be relevant to Akamai.