Monday, March 22, 2010

Mobile Analytics: vertical-specific vs. traditional Web Analytics solutions

Toyota German mobile site(Updated April 19th 2010: following comments, minor updates have been made regarding links & notes)
Earlier this month, the company I worked for launched its first two European sites dedicated to mobile – the Toyota German mobile site and the new Lexus CT 200h pan-European site.

The Toyota German mobile site was specifically designed for high-end mobile devices such as iPhone and provides information on “hot” topics,  Toyota models (pictures, specifications, colours, prices…), a simple configurator and a retailer locator based on Google Maps services.

The Lexus European mobile site supports the launch of new Lexus C-premium hybrid car, providing pictures and information on the car features plus the possibility to register to the newsletter.

Lexus CT 200h minisite for mobile devices

Enough with the promotion… From my perspective, as a Web Analytics freak, this meant my first steps in mobile analytics! Hurray! (me jumping in the air, full of joice and happiness ;-))

Mobile vs. Web analytics
This is an exciting experience as it is opportunity to expand my expertise and learn new things. It also means new challenges to tackle as measuring mobile sites does not work the same way as measuring Websites. The differences come from the specific aspects of the mobile platforms and devices.

The biggest challenges with mobile analytics are
  • Identifying unique visitors because mobile devices do not support cookie and the changing of IP addresses
  • Collecting client information such as device model, manufacturer, screen resolution and others. Most common ways of collecting data for web analytics relies on the use of Javascript but not all mobile devices support execution of Javascript. Gasp!
  • And finally, retrieving geographic information (if important) can be difficult as well – again not all devices enable geographic detection.
This means that traditional “out-of-the box” Web analytics tagging method will not work (or with limitations). Oh damned! So what can we do?

Well there are other methods to collect data. Phew! We are saved. So what are these alternatives? Basically there a four main solution types:
  • Server log-based: process the raw data coming from you web server (My take is to forget about this one except if you are really a geek and that you have plenty of time to loose! :-))
Mobile analytics: log-server based solution
    • Packet sniffing: a hardware device or software is added between your server and the Web. It listens and analyses requests sent and received by the server. No need of having tags here but additional hardware or software is required (meaning you will need help from your IT best friends!)
    Mobile analytics: packet sniffing based solution
      • Image tag-based: A call to an image is added in the content with appropriate parameters & value in the querystring. Data are sent to a collector server by the client device each time a page is rendered. Parameters can be dynamically set by the server back-end when the content is generated. Easy but it has some limitations (see previously) and can not be used for event tracking.
      Mobile analytics: tag-based solution
        • Server-side script: A script is added on server side that sends data directly to collector server when requests are processed. It doesn’t rely on the device to get the data. Data are extracted from the received requests and set dynamically based on the request/content. This solution is getting more and more used by traditional Web Analytics vendors.

            Web Analytics solution or vertical-specific mobile one? 
            Less than a year ago, the best option would probably have been considering a specific mobile analytics solution such as AdMob(1), Bango, Percent Mobile (2), Amethon to name few of them. But mobile analytics has become almost as hot as social media monitoring and it only took a few months to see most Web Analytics vendors offering technical solutions for mobile site tracking whether this is Google Analytics, WebTrends, NedStat, Omniture, AT Internet… Just check with you nearest local vendor for more info.

            So which one should you choose? Before considering a tool, you must first need to clearly identify what are the business objectives of your mobile site and its strategic importance. This will help you defining what KPI’s, metrics & the required level of details you need (see here if you need a good methodology to define your online KPI's). For example, do you need to go as deep as identifying model of a specific device brand or knowing the platform is enough for you? Keep in mind that it is not the quantity of data you can get that matters, it is what you do with it!

            Think wisely!
            My personal opinion is that mobile specific vendors have lost a big part of the edge they had compared to traditional Web Analytics vendors. If possible, you should try first using your existing Web Analytics solution. Why? Here are some points to consider:
            • Tagging & data consistency: using same platform for both your Web and mobile sites means you can tag & measure content in the same way. If you need to compare web vs. mobile performances, at least you will be comparing apples with apples.
            • Web & Mobile data integration: By having both Web and mobile data on same platform – data can be aggregated and segmented at will. For example, what content is more efficient on mobile? It will give you a global view of your online ecosystem on one platform while being able to isolate each main segment.
            • Leverage existing knowledge & expertise: why learning a new tool when you can leverage years of expertise you already have? It can be fun of course to learn new tools but do you have the time? Can you or your business stakeholders afford the risk?
            • Reusability: If your mobile site is an extension of your website (or very similar), you may reuse existing tagging guidelines, mechanisms but also reports and indicators. No need to reinvent the wheel! In my case, for our mobile site, 80% of reports are existing ones (from main site) and 20% are mobile specific ones
            • Costs: If you go for another solution there may be additional costs involved either for the software/hardware or for the technical expertise and implementation.
            Of course, you must make sure that your most important needs are covered and if there are limitations, these are acceptable in your business context.

            Use a Kaizen approach
            If Mobile analytics is new to you (as it is for me), I also recommend a Kaizen approachstart small & simple, use first the tools and expertise that you have. It is a learning process. It is better to take the easy way and improve & develop your mobile analytics capabilities step by step rather than a big leap with the risk to make it more complex and more risky than necessary.

            If you find out it is not enough or too limited – think how you can enhance it easily. For example, you can use a image tag-based implementation using your existing Web Analytics tool and complement it with a free mobile specific solution to get missing information such as used devices & handsets, screen resolution…

            When you reach the limits of your framework then it will be time to consider moving to something else – plenty of choice there – by that time you should have gained enough experience and expertise to make the right choice and justify any additional investments in time, resources and money. Well, that’s what I think. Future will tell if I am right.

            So what do you think? Does it make sense? What are your thoughts on the different methods & solutions? Any other key aspects you would consider before making a choice? Any experience with mobile analytics you are willing to share? I would be curious to know.

            Related articles:
            (1) Note that in November 2009, Google acquired AdMob (more info here) but the deal is under investigation (maybe seen as "anti-competitive").
            (2) Former TigTags analytics solution has been "absorbed" by Percent Mobile.

            Feedback Form