Practical Analytics by DigiKnow

Analytics for iPhone Apps

Usually when I talk about analytics, we are focused on websites or at least apps that live within a browser.  Today, I want to look at analytics within mobile apps, specifically for iPhone apps.  Although we’ve build iPhone apps before, we didn’t care much about analyzing usage previously, but today we launched a new app, the Latin Recipe Shaker, and because type of app it is and the scope of the advertising around it, we wanted to make sure everything we did was measurable.

The first challenge we faced was the decision on which analytics package to use.  We ended up focusing on two main solutions: Flurry and Google Analytics.  In looking at both tools from an analysts perspective, I didn’t see much difference in the reporting provided.  The way the user’s location is tracked seems to be a bit more robust in Flurry, but since the app isn’t mobile aware, that was not a factor for us.

The second question was about implementation.  After having my developers look at both systems, they said the implementation was nearly identical, so there wasn’t any preference on their side.  Both systems use an asynchronous processing of data so that every click and swipe isn’t pushed up in real time.  Flurry uploads the data from the previous visit when the app starts.  Google Analytics lets you set a delay, which defaults to 10 seconds, and then it uploads the data on that interval.  Although it uses more pushes, I prefer the Google Analytics route.

Since all things were equal, and since DigiKnow is a Google Analytics Authorized Consultant, we decided to go with Goggle Analytics.

The implementation basically allows for two types of tracking: page views and events.  From what I know, you can’t do custom variables yet.  We setup page views for each of the main screens and did events for the settings changes, adding/removing on the favorites page and the keyword recipe search. 

The only major hiccup that we’ve seen so far is that the Visits data doesn’t seem to be quite right.  Somehow a single real-life visit shows up as multiple visits in Google Analytics.  This may be related to the asynchronous processing.  The other data point that isn’t quite right is geography.  Since we didn’t want to use the iPhone GPS, our app just uses IP for location, and every iPhone that is on 3G or Edge looks like it is in Dallas (the home of AT&T). 

Overall, the process of implementing analytics was fairly simple and the data coming in is very valuable.

Analytics @ SXSW

I was fortunate to be able to spend four days last weekend down in Austin, TX at South By Southwest’s (SXSW) Interactive Conference.  In addition to meeting a ton of great people from around the world, there were amazing sessions on every topic under the sun, from user experience to mobile, from copyright to managing Gen Y.  One thing that was obviously missing from the conference though was a discussion on analytics.  Of the several hundred panels within the interactive schedule, there was just one panel on analytics.  Today, now that I’m recovered and back in the swing of things, I checked the panel picker, where interactive fans and attendees can vote on which topics will be presented at the conference, and of the 2400 submitted ideas, only 20 had anything to do with analytics (some very loosely).  Analytics is clearly underrepresented.  Next year, I think I’ll need to do my part to try and fix that!

Just One Panel?

The one panel that  made it onto the conference was “What Are Analytics? A Guide To Practical Data”, which had a slightly misleading title, since it was 100% about social media analytics. Margaret Francis and Blake Robinson lead a packed room of over 200 attendees and did a nice overview of the top things to be measuring within social media campaigns.  Some of their key points included a focus on social media sentiment, comparing your brand/company to the competition and considering the tangential relationships that can be found by going down the rabbit hole of tag clouds and keyword usage.  The panelists admitted that we are very much in the infancy of social media measurement.  While tools like PostRank.com, TwitterCounter.com and Nielsen's BlogPulse are useful, they are not at all equivalent to the tools that we have to measure website interactions today.  Geographic and demographic profiles on social media are barely directional at this point, but everything will continue to evolve.  Their key takeaways were: make data visible, set goals and metrics, track performance over time.

Mobile Tracking

As part of the conference, DigiKnow had the opportunity to work on a mobile project with one of the event sponsors (who I can’t mention by name).  I wanted to share a little of this experience, since it was our most focused effort in measuring mobile.  For this site, we initially setup Google Analytics with image based mobile tracking, figuring that would cover both smart phones that could run the JavaScript tracking and feature phones that might not be able to.  On Sunday (half way through the conference), our data seemed low and we wondered if there was a problem, so we added in a second profile running the JavaScript tracking.  As it turned out, the image tracking was only getting about half of the data that the JavaScript tracking revealed.  Today, well after the fact, I also ran the site’s logs through a log-based tool and it came back with data that was very similar to the JavaScript tracking.  Given the fact that the typical SXSW crowd are smartphone users, this wasn’t too surprising.

Key takeaways:

• For mobile, if you can use multiple collection methods (image, JavaScript, logs) you will get the most complete picture, but if you can only do one, I recommend sticking with standard JavaScript tracking

• Any time you are using a new tool or data collection method, test it thoroughly before launching it.  We needed more test data to reveal the deficiencies in the image tracking for mobile in our project.

Measuring Mobile

Two of my favorite aspects of the web world today are analytics (if you hadn’t guessed) and mobile.  So what better thing to do than talk about them both together!  If you aren’t thinking about mobile for your company or your clients, it’s time to start thinking about it.  If you need some compelling data, you can check out my latest mobile marketing presentation on SlideShare.

The first question I get is “Do I need a mobile website?”  And of course as a good analytics guy, my answer is “Let’s look at the data.”  There are several ways to look at the amount of mobile visitors you get on your website.  The easiest is way to see your mobile traffic is to look at the web browsers and/or operating systems used to access your website.  Almost all analytics packages will show you this basic data about your visitors. 

Using the raw browser usage data isn’t super accurate, since 1) it doesn’t easily group all mobile browsers and 2) it includes all visitors, including repeat visitors.  Here’s how I get around this in Google Analytics.

1. Setup a segment in to look for all mobile devices.  I find looking at any device that has a screen smaller than 640 pixels wide is a good measure. 

Setup: Screen Resolution less thank 640

Update: Now that GA updated their segment management, this URL will let you grab my mobile segment – Mobile (Small Screen)

2. Look at the mobile segment as a percentage of new visits, rather than total visits.  Since you don’t have a mobile-friendly site yet, I always assume that people don’t come back as returning visits from their mobile device.  The new visitor data is under Visitors, then New vs. Returning.  With All Visitors and the Mobile segment active, you’ll see the mobile new visitors number as a percentage of total website visits. 

 

 

The real question here is what is a high percentage of mobile visitors?  From my experience mobile, new visitors range from 0.30% up to 2.0% for my clients.  Now that doesn’t compare at all to something like ESPN where they have said that on some weekends they get up to 50% of their traffic from mobile, but let’s face it, we aren’t ESPN.

If over 1% of your new visitor traffic is from mobile, you might want to start considering that audience. 

 

Technorati Tags: mobile,segment,google,smart phone