Mobile analytics is a hot topic for companies seeking to innovate in a new channel for their products, extend their brand, and expose content and advertising to a wider audience. Several Web analytics vendors offer mobile measurement capabilities mixed into their current offerings. Some even offer mobile specific reporting. A few companies provide specific solutions for mobile analytics as part of their ad networks, content delivery and transactional processing systems, marketing and barcoding services, and even as infrastructure or network appliances. Even audience measurement companies have entered this space, primarily through acquisition.
For the most part, when people refer to "mobile analytics" they are referring to collecting behavior about "mobile browser" activity across a variety of handsets. Very few companies offer the ability to track behavior in non-browser-based mobile applications.
Measurement challenges in this area include:
Data collection. Not all mobile browsers execute JavaScript, so the most common method for collecting analytics data doesn't work across all devices. Thus, vendors offer us choices for data collection. Current mobile analytics offerings include image-based data collection methods, packet sniffers, server-side "no tag" implementations, and log files.
Unique visitor identification due to lack of cookie support and the changing of IP addresses. IP addresses on mobile browsers can change as they switch from tower to tower. In addition, many mobile devices will take the IP address of the gateway, making all the devices look the same "person."
Compounding the difficulty in assessing "uniqueness" is that not all mobile devices support cookies. As many of you know, in Web analytics, cookies are helpful in defining uniqueness, and in mobile analytics are helpful in weaving together sessions when the IP address changes mid-session. The fallback method in analytics, when you can't use a cookie, is IP address/user agent. Thus, if you can't set cookies and the IP address and user agents are identical, then how do you identify uniqueness? That's the challenge. Interestingly, packet sniffing as a data collection method has an advantage here because some devices pass unique IDs (such as the phone number) in the HTTP header. When you can detect a unique value in the header, you can easily detect uniqueness.
Handset capability detection. Companies that want to identify whether the device supports WAP pushing, streaming video, ringtones, downloading video clips, and so on need to carefully select a measurement tool in order to ensure these attributes are available.
Phone and manufacturer identification. Databases from WURFL and DeviceAtlas can be used to identify phone and manufacturer device attributes. Larger vendors are further behind on integrating this data into their current offerings, whereas the smaller niche players are making use of it.
Screen resolution detection. The Mobile Marketing Association' s standards for the four "standard" screen sizes may carry enough weight to push this disdained piece of metrics trivia available from JavaScript based tagging in web analytics into a brighter spotlight for guiding user experience and interface design for mobile applications. Traffic source detection. Determining the source of traffic, such as search, email, direct entry, RSS feeds, and marketing campaigns can be challenging in the mobile space.
Geographic identification. Where are the visitors viewing your site coming from? And what does the mobile audience environment "look like" in each country? From this information, you can extrapolate country-specifics for mobile site and application optimization and localization. But not all devices enable geographic detection because the gateway's IP address is used, not a GPS signal. If geo data is important to you, make sure you ask vendors that you are researching how they collect it and what are the limitations. While there are still many challenges in collecting and reporting mobile analytics data, the industry is much further along than we were last year in delivering solutions in this space. Still there's a lot more work that vendors need to do to improve the precision of the data they are collecting and the overall data about the mobile experience that they are reporting.
As you look toward purchasing the best solution for your company's needs, carefully consider the data you need to collect and report for analysis, and judiciously choose the vendor that provides the most appropriate and extensible data collection and reporting capabilities that fit your business goals.
PRINT
COMMENT
SUBSCRIBE
RSS
RSS FEEDS
I'm confused. Are you saying it's ok to drop the WIFI visitors? Are you saying it's ok to only be accurate on a handful of "important" carriers?
I think mobile web site operators would disagree. They want to see all the data. They don't want it to look like one country is getting more traffic than another when it's not true. They don't want to see "Unidentified on 30% of the traffic they are getting.
Maybe you are not understanding the level of missing data that Bango is providing. Feel free to view some details on our blog @ http://www.mobilewebanalytics.net/?p=32.
We have challenged Bango to an impartial third-party side-by-side comparison test, but have not heard from anyone over there.
We do have tests being run by reputable industry experts, and I look forward to seeing some of the results.
Greg
http://ww.mobilytics.net
One clarification: The IP addresses of a mobile connected through a carrier take on the IP address of the carrier gateways - typically one of a handful each carrier allocates. Its a WiFi device that hops IP addresses when it moves between WiFi networks.
Thats why companies like Bango come it. They use their relationships with carriers to get a unique and invariant identity for each and every user on the most popular carriers. By this means they can distinguish between multiple users visiting a site and the same user visiting repeatedly.
Although Bango does not have relationships with every carrier, that have pretty good coverage of the most important to us as a European/US targetting org. On top of the 400 or so mobile carriers round the world, you will also see users coming in via ad-hoc wifi and bluetooth networks, which of course are not identifiable as "carriers" as they are just random internet connections.
The other thing about Mobile is that things change literally hour by hour as different user types (lunchtimers, evening people etc.) have different behaviour.
Thanks for spreading the word. Excellent post!
We've been getting a lot of interest lately in Mobilytics, so indeed it is a hot topic.
I urge companies venturing into Mobile Analytics to run a few products side-by-side on their sites before committing to one. Comparison testing has shown serious deviations between vendors in this space. We have even found conflicting data coming from the same vendor on different reports. We take pride in being the most accurate product available, and encourage potential clients to really dig into a comparison of all solutions. The only thing worse than no data, is wrong data...
Also, I'm glad you mentioned "non-browser-based" mobile applications as well. We are currently beta testing our J2ME java library that provides handset installed application analytics, and are working on our Android, and iPhone libraries.
This is an exciting space, and it's great to get input from traditional web analytics experts like yourself.
Greg Harris, CEO
Mobile Visions, Inc
http://www.mobilytics.net
Excellent post - incredibly worth while read and very complete!
See you at eMetrics DC!
This is an excellent, well researched article covering all the various techniques for gathering mobile metrics. Unique visitors count is particularly important and was the most wanted stat in a survey we ran recently where 80% of mobile site owners said they wanted to know unique visitor count. Packet sniffing is one way of doing this but is not reliable as many operators don't pass the mobile number through the header (think of the data protection issues here!).
Bango provides mobile analytical data on milions of consumers, across 306 networks worldwide (from last month’s data). We have billing relationships with operators which give us an added information to help with identification but we also provide rich data about users and user behavior across every network we measure.
Sarah Keefe, VP Marketing at Bango http://bango.com